In Claim Raven's analysis of 55,070 BVA decisions where the veteran's branch of service was classifiable, Air Force veterans were granted at 34.1%. National Guard veterans were granted at 21.5%. That's a 12.6-percentage-point gap between the highest-grant-rate and lowest-grant-rate branches in the dataset.

I want to be careful with that number before I do anything else with it. Branch of service doesn't change the law. Every veteran at the BVA gets adjudicated under the same statutes, the same regulations, and the same evidentiary framework. A 12.6-point spread, on its face, looks like an indictment of someone, but the data, when you dig into it, mostly tells a story about what each branch puts into a record before the case ever gets to the Board. Different exposure profiles. Different documentation cultures. Different overlap with presumptive lists. Different rates of appeal to the Board in the first place. The gap is real. The reasons are mostly structural. I'm going to walk through what the data shows and where the data stops.

TL;DR

  • Air Force at 34.1%, Marines and Coast Guard tied at 32.1%, Army at 30.7%, Navy at 29.8%, National Guard at 21.5%. Total spread is 12.6 percentage points across 55,070 classifiable cases.
  • The big three (Army, Navy, Air Force) account for 47,849 of the 55,070 classifiable cases. Within the big three, the gap is only 4.3 points (29.8 to 34.1). That's a small structural spread once you're looking at major-branch veterans.
  • Marines run hot on Camp Lejeune, 166 of their top cases land on the Camp Lejeune presumptive at 44.0% grant. That's a branch-specific tailwind that explains some of the Marine numbers.
  • National Guard at 21.5% is the outlier. Title 32 vs Title 10 status, line-of-duty determinations, and fragmented records across federal and state systems all hurt before the case ever reaches the Board.
  • Coast Guard is small (417 cases) but tracks Marine numbers closely. Don't read much into the Coast Guard rate.
  • Branch isn't destiny. The within-branch variance, driven by condition mix, nexus quality, and evidence development, is much larger than the between-branch gap.

The Branch Table

Grant rates by branch

Here's the headline data. Each row reads branch, total classifiable BVA cases in the dataset, grant rate, denial rate, remand rate.

  • Air Force, 8,513 cases. 34.1% grant / 41.6% denial / 24.2% remand.
  • Marines, 6,366 cases. 32.1% / 44.5% / 23.4%.
  • Coast Guard, 417 cases. 32.1% / 41.5% / 26.4%.
  • Army, 29,227 cases. 30.7% / 44.1% / 25.3%.
  • Navy, 10,109 cases. 29.8% / 44.1% / 26.1%.
  • National Guard, 438 cases. 21.5% grant / 47.0% denial / 31.5% remand.

The Board-wide grant rate across all 101,518 condition records drawn from the analyzed subset of Claim Raven's 501,000+ Board-decision library in the broader dataset is 30.6%. So Air Force, Marines, Coast Guard, and Army are all above average. Navy lands just below. National Guard sits 9.1 points below the system mean.

That table is the entire dataset on this question. Everything else in the post is trying to explain what's driving each row.

Why Air Force Lands at the Top

Air Force veterans grant at 34.1%, the highest rate among the major branches by a 3.4-point margin over the closest competitor (Marines at 32.1%, where the sample is smaller). I want to lay out what I think is happening here and then immediately flag what I can't prove.

Air Force veterans in the Vietnam era have the strongest single tilt toward presumptive service connection in the entire dataset. Look at their top-five conditions: 352 COPD cases (35.8% grant), 248 sinusitis cases (31.0%), 236 GERD cases (34.7%), 213 eczema cases (46.9%), and 189 prostate cancer cases (60.8% grant rate, the highest single-condition grant rate among Air Force top-fives).

Prostate cancer at 60.8% isn't an Air Force anomaly. It's an Agent Orange presumptive condition under 38 CFR § 3.309(e), and Air Force veterans served in Vietnam in roles that exposed them to herbicide. The combination of a high-grant-rate presumptive condition and a meaningful Air Force claim volume on that condition tilts the branch number upward.

That same pattern shows up in other Air Force cuts. Air Force veterans had heavier exposure to documented environmental hazards, flight line operations, JP-4 fuel, asbestos in older airframes, ionizing radiation around sensitive equipment. Many of those exposures map onto presumptive lists or onto conditions with strong nexus literature. Burn-pit-era Air Force veterans (Iraq and Afghanistan) appear in the dataset too, with respiratory conditions that benefit from PACT Act presumptions enacted in 2022.

There's also a documentation effect. Air Force bases tend to be larger, more permanent, and better resourced medically. Service treatment records are, on average, more complete and more clinically detailed than at smaller installations. When a veteran later claims a condition that traces to in-service care, an Air Force record is more likely to contain the entry that supports the claim. That's not because Air Force veterans are healthier or sicker than veterans in other branches. It's because the system that documented their care was more standardized.

And there's a demographic skew worth naming. Air Force has historically had a higher share of technical, clerical, and aviation-support MOSs than infantry-heavy branches like Army or Marines. Some of those MOSs map onto conditions that the rating schedule treats favorably (presumptive respiratory conditions, occupational hearing loss with strong audiograms, sleep apnea where command climate during deployment is well-documented). The Air Force claim mix is structurally tilted toward higher-grant-rate conditions.

I want to be honest that I can't fully decompose how much of the 34.1% Air Force rate is "more presumptive conditions in the claim mix" versus "better documentation in the service record" versus "structural exposure-MOS overlap." The data lets me see the conditions Air Force veterans claim. It doesn't let me see all the upstream reasons those conditions are weighted differently in their claim mix than in other branches.

What I can tell you is that the 34.1% isn't randomness. It's a branch whose claim mix and documentation pattern both happen to align with what the Board treats favorably.

Marines, Camp Lejeune, and the Presumptive Tilt

Marines grant at 32.1% on 6,366 cases. That's 1.5 points above the Board-wide average and 2 points below Air Force. I want to spend a moment on the Marine number because it has a specific structural feature most veterans don't immediately see.

Look at the Marine top conditions. TBI at 249 cases (33.7% grant), COPD at 177 (35.6%), GERD at 168 (41.1%), Camp Lejeune at 166 (44.0%), and sinusitis at 159 (31.4%). Camp Lejeune is in the Marine top-five. It's not in any other branch's top-five.

Why that matters: Camp Lejeune is a presumptive-service-connection program under 38 CFR § 3.307(a)(7) and 38 USC § 1710(e)(1)(F). Veterans who served at Camp Lejeune between August 1, 1953 and December 31, 1987 are presumed to have been exposed to contaminated water, and a specific list of conditions is presumptive for that exposure. Bladder cancer, breast cancer, kidney cancer, liver cancer, multiple myeloma, non-Hodgkin lymphoma, and Parkinson's disease are all on the presumptive list. Camp Lejeune was a Marine Corps base, so the veteran population eligible for this presumption is overwhelmingly Marine.

That 166-case Camp Lejeune cluster, granting at 44.0%, is functionally a Marine-specific tailwind. The same is true to a lesser degree for the Marine combat MOSs that overlap with burn-pit-era exposure under PACT Act presumptions.

Strip out the Camp Lejeune and burn-pit presumptive tailwinds and the Marine grant rate would probably converge with the Army-Navy band around 30-31%. I can't run that decomposition cleanly from the data I have, but the directional read is straightforward: Marines benefit from a presumptive cluster that other branches don't.

The other interesting thing about the Marine numbers is the TBI count. 249 TBI claims is the largest top-five appearance of TBI in any branch's data. Marines are a smaller branch than Army, but TBI shows up in their top-five and not in the top-five of any other branch except combat-arms-heavy slices. That's consistent with what we know about Marine deployment patterns, heavy combat-arms presence, exposure to IEDs and blast events, and the post-2004 expansion of TBI recognition under VA's polytrauma guidance.

TBI at the Board grants at 33.7% for Marines, which is essentially the Marine average. The presence of TBI in the Marine claim mix doesn't tilt the rate up or down in a major way. What it does is signal that Marine veterans are claiming conditions that align with their actual exposure profile, which is a sign of a claim ecosystem that knows what to file.

Army and Navy: The Big-Number Middle

Army (29,227 cases) and Navy (10,109 cases) together account for 71% of the classifiable branch data. They grant at 30.7% and 29.8% respectively, within a point of each other and within a point of the Board-wide average of 30.6%. These are the branches where the data is most robust and where the branch effect on grant rate is smallest.

The Army claim mix in the top five is COPD (958 cases, 32.8% grant), TBI (935, 27.3%), sinusitis (715, 29.2%), GERD (661, 36.9%), and eczema (607, 41.8%). That's a profile that looks like the BVA-wide condition mix. Big chronic conditions, big exposure-related conditions, a meaningful TBI presence reflecting Army's role as the largest combat-arms branch. The Army number is essentially the average because Army is essentially the average, the branch is large enough and varied enough that its claim mix mirrors the overall Board docket.

Navy's top five is COPD (599, 28.4%), GERD (250, 33.6%), sinusitis (243, 34.2%), eczema (240, 40.0%), and adjustment disorder (220, 44.1%). The Navy number runs 1.7 points below the Board average. A few structural factors might be doing work there.

Navy claim populations skew slightly older on average, many sailors retired or separated decades ago, and a meaningful share of Navy claims at the BVA involve conditions claimed long after service. The older the gap between service and claim, the harder the nexus development tends to be, and nexus is where the system loses most claims. Navy's slightly lower rate may reflect a slightly older average gap between service and claim, not anything specific to the Navy as a branch.

Navy's adjustment disorder cluster at 220 cases granting at 44.1% is worth flagging. Adjustment disorders generally grant at higher rates than PTSD across the dataset (I've written about this elsewhere), and Navy's adjustment-disorder volume might reflect a service culture where stress-driven mental health claims get diagnosed as adjustment disorder more often than as PTSD. That's a hypothesis. The data can't confirm it.

What I can say for the Army-Navy band is that the 0.9-point gap between them isn't large enough to mean much. If the Board's overall grant rate is 30.6%, both Army and Navy are operating within normal variance around that mean. The interesting thing isn't the difference between Army and Navy. It's that both branches sit close to the Board-wide average, which is what you'd expect from branches whose claim mixes and exposure profiles are big enough and varied enough to look like the average.

Coast Guard's 32.1% on 417 Cases

Coast Guard appears in the data at 417 cases with a 32.1% grant rate. That's tied with Marines for the second-highest grant rate in the dataset.

I want to flag this carefully. 417 cases is the smallest sample among the major branches by an order of magnitude. The 32.1% rate is real but it's sitting on a base where a few dozen case-level fluctuations could shift the rate by a percentage point. I would not draw strong conclusions from the Coast Guard rate.

What I can say structurally is that the Coast Guard claim profile resembles the Navy more than the Army. Top conditions are COPD, GERD, sinusitis, and adjustment disorder. The branch's exposure profile is maritime, with documented hazards around fuel handling, sea-state-related musculoskeletal claims, and a mental-health claim mix that looks like a smaller version of Navy's. The grant rate is essentially what you'd expect from a Navy-like claim mix run through a sample so small that any individual exposure cluster could shift the average.

If the Coast Guard sample doubled to 800 or 1,000 cases, I'd expect the rate to converge back toward the Navy band of 29-30%. The 32.1% is interesting but the sample isn't telling you much.

National Guard at 21.5%. The 9-Point Anomaly

This is the row that needs the most careful treatment.

National Guard veterans in our dataset (438 cases) grant at 21.5%. That's 9.1 points below the Board-wide average and 12.6 points below the top branch. It's also the only branch where the headline grant rate sits clearly below the Board average rather than within normal variance of it.

I want to be explicit that this isn't about the quality of National Guard service. The Guard contributes operationally to nearly every theater the active component deploys to. The 21.5% number isn't a verdict on the Guard. It's a verdict on how Guard claims get developed and adjudicated in the system that produces BVA appeals.

A few structural reasons appear to drive the gap.

Title 32 versus Title 10 service. Guard veterans serve on a mix of Title 32 (state) and Title 10 (federal) status. Most VA disability claims require qualifying service under Title 10, federal active duty. Periods of Title 32 service generally aren't qualifying for federal VA compensation unless the claim arises from an injury or disease incurred or aggravated in the line of duty during Title 32 service that meets specific criteria. The eligibility framework for Guard claims is structurally more complex than for active-component veterans, and that complexity translates into more denials at the regional office and more procedural issues at the Board.

Line-of-duty determinations. Guard veterans claiming injuries from inactive-duty-training periods (drill weekends) or annual training periods need a line-of-duty determination establishing that the injury occurred during qualifying service. Line-of-duty determinations are often not issued contemporaneously, which means decades after the fact, the Guard veteran is trying to establish through reconstructed records that an injury occurred during a specific drill period. That's a heavy evidentiary lift that active-component veterans don't usually face.

Fragmented medical records. Guard medical records live across federal and state systems. A Guard veteran's records may be split between the National Guard Bureau, the state adjutant general's office, the Defense Personnel Records Information Retrieval System (DPRIS), and various unit-level files. Reconstructing a complete service treatment record for a Guard claim is harder than for an active-component claim where the records flow through a single chain. The fragmentation predates VA's involvement and shows up as gaps that hurt at every stage of claim adjudication.

Selection bias at the BVA level. Guard claims that make it to the Board are not a random sample of all Guard claims. They're the claims that got denied at the regional office and that the veteran was willing to appeal. The selection process at the RO may be even harsher on Guard claims because of the eligibility complications I just described, which means the Guard cases that survive to the Board may be on average harder cases than the active-component cases that get there. The 21.5% rate is among Guard appellants, not among all Guard claimants.

Sample size caveat. 438 cases is enough to identify a clear pattern but not enough to decompose every contributor. The Guard rate is 9.1 points below the Board mean. That's statistically real on 438 cases. The decomposition into Title 32 issues, line-of-duty issues, records fragmentation, and selection effects is qualitative, based on reading enough Guard decisions to see the pattern, not a quantitative breakdown I can produce from this data.

What I want to be clear about is that the 21.5% isn't an indictment of Guard veterans or of Guard service. It's a description of a structural disadvantage that Guard claims face in the system that adjudicates them. The system's design assumes a uniform active-duty service period with continuous federal records. Guard service doesn't fit that assumption cleanly, and the consequences show up at the BVA in the form of more denials and more remands than active-component claims face.

If you're a Guard veteran reading this, the 21.5% is not a prediction for your specific claim. It's an average across a small dataset of Board appeals, weighted toward the harder cases by definition. Your individual outcome depends much more on the quality of your line-of-duty documentation, the completeness of your reconstructed service treatment records, and the specificity of your nexus opinion than on the branch average.

Branch × Top Condition: Where Each Service's Story Lives

How condition mix drives the spread

The branch numbers don't tell you everything. The condition mix within each branch is where the real story lives. Let me walk through the top-condition cuts and what they say about each service's claim pattern.

Army top five:

  • COPD: 958 cases, 32.8% grant
  • TBI: 935 cases, 27.3% grant
  • Sinusitis: 715 cases, 29.2% grant
  • GERD: 661 cases, 36.9% grant
  • Eczema: 607 cases, 41.8% grant

This profile reads as the BVA average with an Army-specific TBI tilt. Army has the largest combat-arms population among the branches, and TBI in the top five at 935 cases is consistent with that. The TBI grant rate of 27.3% sits below the Army average, which reflects what we know structurally about TBI claims, they're nexus-hard, often involving subtle cognitive and behavioral symptoms that don't map cleanly onto the rating schedule, and they tend to need specialized neurological documentation that the standard C&P examiner may not produce.

Air Force top five:

  • COPD: 352 cases, 35.8% grant
  • Sinusitis: 248 cases, 31.0% grant
  • GERD: 236 cases, 34.7% grant
  • Eczema: 213 cases, 46.9% grant
  • Prostate cancer: 189 cases, 60.8% grant

The prostate cancer line is the single most distinctive Air Force feature. 60.8% on 189 cases is the highest single-condition grant rate among any branch's top-five appearance of a major condition. That's the Agent Orange presumptive doing its job. Vietnam-era Air Force veterans with prostate cancer have a presumptive path to service connection that doesn't require the nexus development that drags down direct-connection claims. Eczema at 46.9% is also above the average, and the burn-pit presumptive under PACT Act gives some of those eczema claims a structural lift.

Navy top five:

  • COPD: 599 cases, 28.4% grant
  • GERD: 250 cases, 33.6% grant
  • Sinusitis: 243 cases, 34.2% grant
  • Eczema: 240 cases, 40.0% grant
  • Adjustment disorder: 220 cases, 44.1% grant

Navy's condition mix is the most generic-looking in the data. Big chronic conditions, no single dominant tilt. The adjustment disorder rate at 44.1% is notable, adjustment disorders generally grant at higher rates than PTSD across the system, and Navy's adjustment-disorder cluster is consistent with that. The lack of a strong branch-specific presumptive cluster (no Camp Lejeune equivalent, less Vietnam-era Agent Orange overlap than Army or Air Force) shows up as a Navy rate that sits right at the Board average.

Marines top five:

  • TBI: 249 cases, 33.7% grant
  • COPD: 177 cases, 35.6% grant
  • GERD: 168 cases, 41.1% grant
  • Camp Lejeune: 166 cases, 44.0% grant
  • Sinusitis: 159 cases, 31.4% grant

I've already talked about Camp Lejeune. The Marine TBI cluster at 249 cases (more than 4x the Air Force TBI volume on a smaller total caseload) is the Marine signature. So is the slightly elevated GERD rate at 41.1%. Marines may be claiming GERD secondary to PTSD or to combat-related stressors at higher rates than other branches, which boosts the GERD grant rate via the secondary-connection path.

National Guard top five:

  • Sinusitis: 26 cases, 26.9% grant
  • Eczema: 18 cases, 33.3%
  • COPD: 14 cases, 21.4%
  • Migraines: 13 cases, 46.2%
  • GERD: 13 cases, 15.4%

The Guard sample is so small at the condition level that individual case shifts move the percentage by 5-10 points. I want to be careful not to read too much into any specific Guard condition rate. What's noticeable is that the conditions Guard veterans bring to the Board look more generic-respiratory and generic-musculoskeletal than the conditions other branches bring. There's no presumptive-list cluster in the Guard top five. No PACT Act-eligible burn-pit cluster of size. No Camp Lejeune analog. Guard veterans, when they appeal to the Board, are appealing claims that mostly need standard Shedden-three-prong development. The system is harsher on those claims than on presumptive ones, which explains some of the Guard rate.

Coast Guard top five:

  • COPD: 19 cases, 26.3% grant
  • GERD: 16 cases, 18.8%
  • Sinusitis: 14 cases, 21.4%
  • Adjustment disorder: 11 cases, 63.6%

Same caveat as Guard. Small samples. The 63.6% adjustment-disorder rate on 11 Coast Guard cases is interesting but not interpretable as a Coast Guard pattern, that's a number that could move 15 points either way if a handful of cases changed.

What Branch Tells You About Claim Type, Not Claim Outcome

I want to land an important framing. The 12.6-point spread across branches isn't telling you that branch of service determines BVA grant rate. It's telling you that branch of service strongly predicts what kind of claim a veteran files, and the kind of claim is what determines the grant rate.

Air Force veterans file more Agent Orange presumptive claims because more Air Force veterans served in Vietnam roles that exposed them to herbicide. Marine veterans file more Camp Lejeune presumptive claims because Camp Lejeune was a Marine base. Army veterans file more TBI claims because Army has the largest combat-arms population. National Guard veterans file more standard Shedden-three-prong claims because they lack the presumptive-cluster overlap that lifts the branches with concentrated exposure histories.

When you hold the condition constant, when you look at, say, COPD claims across branches, the branch effect on grant rate shrinks substantially. COPD grant rates by branch run from 21.4% (Guard, 14 cases) to 35.8% (Air Force, 352 cases). The bulk of branches sit in the 26-36% band on COPD. That's a 10-point spread on a single condition, but it's much smaller than the overall branch spread when condition mix is allowed to vary.

The interpretation isn't "Air Force veterans win more because Air Force is somehow privileged." The interpretation is "Air Force claim mix happens to include high-grant-rate presumptive conditions in larger proportions than other branches." Those are different statements.

This matters for the way you think about your own claim. Your branch isn't a fate. The condition you're claiming, the connection theory you're using, and the evidence you've developed are what determine your odds. If you're an Army veteran and you have a presumptive condition on the burn-pit or Agent Orange list, you have access to the same presumptive grant-rate tailwind that an Air Force veteran does. If you're a Marine veteran and you don't have a Camp Lejeune-eligible service period, the Marine Camp Lejeune tailwind doesn't help you. If you're a Guard veteran with strong line-of-duty documentation and a complete reconstructed service treatment record, your claim is competing on the same terms as an active-component claim.

The branch effect is structural. The within-branch effect, driven by condition selection, evidence development, and nexus quality, is much larger and is the part you control.

What I Can't Tell You From This Data

A few things this post cannot address.

I cannot tell you what the regional office grant rates by branch look like. This dataset is BVA-level. Veterans whose claims were granted at the RO and who never had to appeal are not in this dataset. Branch effects at the RO may be larger or smaller than the branch effects I'm describing. The 12.6-point spread is among BVA appellants, not among all claimants.

I cannot tell you whether the branch differences reflect actual disability incidence or just claim-filing patterns. Air Force veterans may file claims more aggressively than Guard veterans, regardless of underlying disability rates. Conversely, Air Force veterans may have lower true disability incidence but higher grant rates because their claims are better-developed. The data can't separate "more claims filed" from "more claims granted per filed" cleanly at the branch level.

I cannot tell you the within-branch breakdown by MOS or by service era. Those are separate cuts I've covered in other posts. The branch table aggregates across all MOSs and all service eras for each branch, which means the Air Force number mixes Cold War radar techs with Vietnam-era pilots and post-9/11 cyber operators. Each of those sub-populations has its own claim profile, and the branch average obscures that variance.

I cannot tell you whether the National Guard rate would change if the sample were larger. 438 cases is enough to identify the pattern but not enough to do robust subgroup analysis within the Guard population. The Guard rate could converge upward toward the active-component band as the dataset grows. It could also stay where it is or move further away. The data I have can't predict that direction.

I cannot tell you whether the branch effect is stable across years. The dataset overrepresents recent decisions (2024-2025) because BVA decisions are released over time and the extraction date matters. If the PACT Act presumptive ramp-up has shifted branch outcomes year-over-year (likely tilting Air Force and Marines higher in recent years due to burn-pit and Camp Lejeune effects), the branch table reflects that recent skew. Earlier-cohort outcomes may look different.

I cannot tell you whether any branch is being adjudicated more or less fairly than any other. The variance I describe is consistent with structural differences in claim mix and documentation quality, but the data can't rule out subtle adjudicator effects at the regional office or Board level that might disadvantage one branch over another. I don't see evidence of that in the numbers, but absence of evidence isn't evidence of absence.

What This Means For Your Claim

If you're a veteran reading this and trying to figure out what your branch means for your claim, here's the honest answer.

Your branch is not a meaningful predictor of your individual outcome. The branch averages reflect aggregate claim-mix patterns. Your claim has its own facts, a specific condition, a specific service period, a specific exposure history, a specific nexus situation. Those facts matter far more than which service patch your DD-214 carries.

If your branch has a structural tailwind that applies to your specific case. Marine Corps service at Camp Lejeune during the contaminated water years, Air Force or Army service in Vietnam with documented herbicide exposure, any service in a PACT Act burn-pit-covered location, use that tailwind. File presumptively under the appropriate regulation. Cite the statute. Reference the exposure category in your claim narrative.

If your branch doesn't have a tailwind that applies to your situation, including most active-component peacetime claims and most Guard claims, the variables you control still matter more than the branch average. Build the nexus opinion. Get the C&P exam right (or supplement it with a private medical opinion when the VA exam is inadequate). Write specific, dated lay statements. The branch average is a description of an aggregate. Your file is what gets adjudicated.

Guard veterans in particular should not read the 21.5% number as fate. The Guard average reflects structural disadvantages in how Guard claims get documented and adjudicated, not a verdict on Guard veterans. If your line-of-duty documentation is solid, your reconstructed service treatment record is complete, and your nexus opinion is strong, your individual claim is competing on the same terms as any other veteran's. The Guard structural disadvantage shows up as a population average. Individual claims can clear it.

The 12.6-point spread is a fact about the system. The response to it is to make your file as strong as possible so the average matters less.

Bottom Line

Air Force veterans grant at 34.1% at the BVA. National Guard veterans grant at 21.5%. The 12.6-point spread is real, statistically meaningful on 55,070 classifiable cases, and structurally driven by differences in claim mix rather than by branch-specific adjudication. Air Force benefits from Agent Orange and burn-pit presumptive overlap. Marines benefit from a Camp Lejeune cluster that doesn't appear in any other branch. Army and Navy sit close to the system mean because their claim mixes mirror the overall Board docket. National Guard runs 9.1 points below the mean because of Title 32 eligibility complications, line-of-duty determination issues, fragmented records across federal and state systems, and selection effects at the appeal level. None of this is destiny for your specific claim. The within-branch variance, driven by condition selection, connection theory, and evidence development, is much larger than the between-branch spread. Your branch doesn't decide your claim. The completeness of your record does.

Methodology and Limitations

  • Data source: Claim Raven's analysis of 101,518 condition records drawn from the analyzed subset of Claim Raven's 501,000+ Board-decision library from the production case_analysis dataset. The branch-of-service field was classifiable in 55,070 of those records. The remaining 46,448 records either lacked a clear branch indicator in the decision text or had ambiguous classifications that I excluded from the branch table to avoid noise.
  • Sample size: Army 29,227 cases, Navy 10,109, Air Force 8,513, Marines 6,366, National Guard 438, Coast Guard 417. Total 55,070. The Guard and Coast Guard samples are small enough that individual case-level fluctuations meaningfully move the rate.
  • Classification approach: Branch was extracted from BVA decision text references to military service. When a decision referenced multiple branches (e.g., reservist with prior active service), the primary branch of qualifying service was used. Cases with ambiguous branch references were excluded.
  • Limitations:
  • These are BVA-level cases only. Veterans whose claims were granted at the regional office and who never appealed are not in this dataset. RO-level branch effects may be different from BVA-level effects.
  • The branch table aggregates across all conditions, all service eras, all MOSs, and all docket types. Within-branch variance on these dimensions is substantial and the branch average obscures it.
  • The National Guard sample of 438 cases is small enough that the 21.5% rate has wider confidence intervals than the larger-branch rates. The pattern is consistent enough across multiple cuts to identify a structural disadvantage, but the precise size of that disadvantage is sensitive to sample composition.
  • The dataset overrepresents 2024-2025 decisions. If branch effects have shifted over time (likely true for PACT Act-eligible branches), the table reflects recent skew more than long-run averages.
  • Selection bias at the appeal level means BVA-rate branch comparisons reflect appellant populations, not full claimant populations. The branches with higher RO grant rates may have systematically different cases reaching the BVA than branches with lower RO grant rates.
  • I cannot fully separate the independent effect of branch from the conditions claimed by each branch. The branch table is a marginal effect, not a controlled comparison. Holding condition constant would shrink the branch spread substantially, but I don't have enough sample in the smaller branches to run that decomposition cleanly.
  • Aggregate patterns are not predictions for individual claims. Your branch is a population-level average. Your file is what gets decided.
  • The framing of structural reasons for the National Guard rate is based on reading enough Guard BVA decisions to identify recurring patterns (Title 32 issues, line-of-duty determinations, records fragmentation). It's qualitative, not a quantitative decomposition of how much each factor contributes.

Disclaimer

I'm not accredited by VA, not a lawyer, not a VSO. This is data analysis, not claim advice. These are patterns from cases that made it to the BVA, they don't predict individual outcomes. If you need help with your claim, work with an accredited representative.

Where to go next

-Landon Founder, Claim Raven | U.S. Army Veteran