Presumptive service connection wins 46.9% of the time at the Board of Veterans' Appeals. Direct wins 41.2%. Aggravation wins 41.3%. Secondary wins 35.1%.

Across 69,078 BVA cases in my dataset where I could cleanly classify the connection theory, the spread between the highest path (presumptive) and the lowest (secondary) is 11.8 points. That's not random. That's the structural difference between four ways of proving a claim, each with its own regulation, its own evidence burden, and its own characteristic failure mode.

If you're filing a VA disability claim, the first decision you make, explicitly or implicitly, is which connection theory you're filing under. Most veterans don't know they're making the decision. The form asks "describe your disability and how it's related to service." You write something. The RO classifies your theory based on what you wrote. The classification determines what you have to prove.

I want to walk through the four paths, what each one requires under the regulations, where each one wins and loses in the data, and how to choose the right path for your case. This is the regulatory map that sits underneath every grant, denial, and remand at the Board.

TL;DR

  • Four connection paths in the data: presumptive (8,975 cases, 46.9% grant), aggravation (886 cases, 41.3%), direct (40,026 cases, 41.2%), secondary (19,189 cases, 35.1%).
  • Presumptive wins highest because the regulations remove the nexus burden, no medical opinion required when criteria are met.
  • Direct is the default path: three-prong Shedden test, used in 58% of classified cases.
  • Secondary requires a service-connected primary condition first. Lower grant rate because the proof burden compounds.
  • Aggravation requires "clear and unmistakable evidence" of in-service worsening, a heightened standard.
  • The path you pick isn't always optional. Sometimes the facts of your service determine which path applies. But where you have a choice, the data points to presumptive first, then direct, then secondary.
What evidence wins a VA claimFree

Strong medical opinions won 89.7% of the time in our analysis. Weak ones won just 3.7%.

  • Know what makes a medical opinion harder to dismiss.
  • Use your own statement where it can carry real weight.
  • Catch common C&P exam problems before they hurt your claim.
Get the free guide

The Four Paths In One Chart

Grant rates by connection type

Here are the connection-type grant rates from the 69,078 BVA cases where I could classify the theory.

  • Presumptive: 8,975 cases, 46.9% granted, 25.2% denied, balance remanded.
  • Aggravation: 886 cases, 41.3% granted, 24.3% denied.
  • Direct: 40,026 cases, 41.2% granted, 31.7% denied.
  • Secondary: 19,189 cases, 35.1% granted, 22.8% denied.

The "extraschedular" category appears twice in the data and is anomalous; I'm excluding it.

Three observations before I go into each path individually.

First, presumptive sits 5.7 points above direct and 11.8 points above secondary. That's the biggest within-data spread of any classification I look at, bigger than the branch-of-service spread, bigger than the era spread, bigger than the nexus-quality spread within most condition groups. The connection theory is doing real work in outcomes.

Second, the denial rates tell a different story than the grant rates. Direct has the highest denial rate (31.7%), meaning when direct claims lose, they tend to lose cleanly. Aggravation has the lowest denial rate (24.3%) but a much higher remand rate, which signals the Board frequently can't resolve aggravation claims on the existing record and sends them back. Secondary has a low denial rate too, but its case-level "win or hold" rate (grant plus remand) is lower than the other three.

Third, the case counts are wildly unequal. Direct dominates at 40,026 cases. Presumptive is around 9,000. Secondary is around 19,000. Aggravation is tiny at 886. That's because aggravation is the path with the narrowest applicability, you only claim aggravation when you had a pre-existing condition at entrance that worsened during service. Most veterans don't have a documented pre-existing condition.

Path One: Direct Service Connection

Regulatory basis: 38 CFR § 3.303. Grant rate at BVA: 41.2% across 40,026 cases. Use case: Condition originated during service, or is caused by service.

Direct service connection is the default. When a veteran files a claim and doesn't explicitly invoke another theory, the claim is processed as a direct service connection claim under § 3.303.

The three prongs come from Shedden v. Principi. You can look up Shedden and related precedents in the CAVC case reference. Summarized:

  • Prong one: a current disability.
  • Prong two: an in-service event, injury, or disease.
  • Prong three: a medical nexus linking the in-service event to the current disability.

Most direct claims that fail at the BVA fail on prong three. The veteran has the diagnosis (prong one). The service records show the in-service event (prong two). The medical opinion linking the two is either missing, weak, or contested.

Direct grants at 41.2% across the dataset because direct is the path that requires the most external proof. The veteran has to come with a nexus opinion. The VA doesn't owe one unless McLendon v. Nicholson is triggered, and even when McLendon triggers a C&P exam, that exam can produce a negative opinion that the veteran then has to overcome with private evidence.

What direct does well:

  • It's available for any condition. There's no list of qualifying conditions, no exposure requirement, no service-era requirement.
  • It scales with evidence quality. Strong nexus opinions push direct grant rates above 89% in my data. The path rewards good evidence directly.

What direct doesn't do well:

  • It puts the full nexus burden on the veteran. If you don't have a private nexus letter and the VA examiner's opinion is negative, direct is an uphill fight.
  • The "continuity of symptoms" alternative under § 3.303(b), where you can prove direct connection by showing continuous symptoms from service to present, is restricted by Walker v. Shinseki to conditions on the § 3.309(a) chronic-disease list. For most conditions, you can't ride continuity alone.

When direct is the right path:

  • Your condition is not on a presumptive list, or you didn't serve in a covered area for a presumption.
  • Your condition isn't downstream of an already service-connected condition.
  • You have, or can develop, a medical opinion linking your condition to a specific in-service event.

Path Two: Presumptive Service Connection

Regulatory basis: 38 CFR § 3.307 (general presumption rules) and § 3.309 (lists of specific presumptive conditions). Also various PACT Act expansions, Camp Lejeune-specific provisions, and exposure-specific listings. Grant rate at BVA: 46.9% across 8,975 cases. Use case: Condition is on a specific presumptive list, AND the veteran's service meets the exposure or era criteria for that presumption.

Presumptive service connection is the regulatory shortcut. Instead of requiring the veteran to prove that a specific exposure or service event caused a specific condition, the regulations presume the causal link for defined combinations of exposure-and-condition.

The major presumptive categories in my data are:

  • Agent Orange (38 CFR § 3.309(e)), herbicide-exposed veterans who served in Vietnam, the Korean DMZ during specified periods, certain Thai air bases, and other covered locations. The presumptive condition list includes Parkinson's disease, multiple cancers (prostate, lung, larynx, soft tissue sarcoma, etc.), ischemic heart disease, type 2 diabetes, and other conditions.
  • Camp Lejeune (38 CFR § 3.307(a)(7)), veterans who served at Camp Lejeune between August 1953 and December 1987 with exposure to contaminated water. Presumptive conditions include kidney cancer, liver cancer, Parkinson's disease, adult leukemia, multiple myeloma, and others.
  • Gulf War (38 USC § 1117, 38 CFR § 3.317), veterans who served in the Southwest Asia theater of operations from August 2, 1990 onward. The Gulf War presumption covers undiagnosed illnesses and a set of "chronic multisymptom illnesses" including fibromyalgia, chronic fatigue syndrome, and functional gastrointestinal disorders.
  • Burn pit and PACT Act expansions (PL 117-168), veterans exposed to airborne hazards from open burn pits in Iraq, Afghanistan, and other covered locations. The PACT Act added 23 new presumptive conditions in 2022, including several cancers, respiratory conditions, and chronic skin conditions.
  • Radiation exposure (38 CFR § 3.309(d)), atomic veterans and others with radiation exposure under defined criteria.
  • POW status (38 CFR § 3.309(c)), former prisoners of war, with a separate presumptive condition list.

When all the criteria for a presumption are met, the right exposure or service in the right location during the right period, the right diagnosed condition, the veteran does not need a nexus opinion. The connection is presumed.

That's why presumptive grants at 46.9%. The single biggest reason claims fail at the BVA, missing or weak nexus opinion, is structurally removed from presumptive cases. The proof burden shifts from "did service cause this condition?" to "did the veteran meet the criteria for the presumption?"

What presumptive does well:

  • It removes the nexus requirement. No medical opinion linking exposure to condition needed.
  • It's binary on eligibility. You either meet the exposure criteria or you don't. Once you do, the connection is automatic.
  • It scales with policy expansion. Every PACT Act addition and every new presumptive listing expands what's available.

What presumptive doesn't do well:

  • It's narrow. If your condition isn't on a presumptive list, the presumption doesn't apply.
  • It still requires exposure proof. "I served in Vietnam" is the trigger for Agent Orange presumption only if the veteran actually served in Vietnam during the covered period. Boots-on-ground proof, service in a covered location during a covered period, or other documented exposure has to be established.
  • Diagnosis still has to be confirmed. The presumption assumes the link between exposure and a specific condition. It doesn't presume the diagnosis itself. A veteran who claims Parkinson's under Agent Orange presumption still has to have Parkinson's confirmed.

When presumptive is the right path:

  • Your condition is on a presumptive list, AND
  • Your service meets the exposure or location criteria for that presumption.

If both are true, presumptive should be the primary theory of the claim. Always.

Path Three: Secondary Service Connection

Regulatory basis: 38 CFR § 3.310. Grant rate at BVA: 35.1% across 19,189 cases. Use case: Condition is caused or aggravated by an already service-connected condition.

Secondary service connection is the path for conditions that develop downstream of a primary service-connected condition.

The classic chains:

  • PTSD → sleep apnea (the trauma → sleep disturbance → apnea pathway)
  • PTSD → hypertension (chronic stress → cardiovascular effects)
  • Knee injury → back problems (altered gait → spinal degeneration)
  • Diabetes → kidney disease (diabetic nephropathy)
  • Diabetes → peripheral neuropathy (diabetic nerve damage)
  • Back injury → radiculopathy (nerve root compression secondary to spinal pathology)

The proof structure for secondary under § 3.310 is:

  • The veteran has a current secondary disability.
  • The primary condition is already service-connected.
  • A medical nexus links the secondary condition to the primary condition.

The aggravation prong under § 3.310(b), sometimes called "secondary aggravation", is a variant where the primary condition aggravates a non-service-connected condition. The same proof structure applies, but the rating runs to the degree of aggravation rather than the full disability.

Why does secondary grant at 35.1%, lower than direct (41.2%) and presumptive (46.9%)?

Two structural reasons.

The first is that the proof burden compounds. To win a secondary claim, you have to prove the secondary nexus and the primary condition has to already be service-connected. If the primary isn't yet service-connected, you have to prove that too, usually in the same claim or appeal. Two nexus questions instead of one.

The second is that the medical literature linking secondary conditions to primary conditions varies in quality. Some links are clean and well-established, diabetes to nephropathy, knee injury to compensatory back strain. Others are more contested. PTSD to hypertension, PTSD to sleep apnea (which has been the subject of significant litigation). When the medical literature is contested, the nexus opinion battle gets harder.

What secondary does well:

  • It opens up compensation for downstream conditions that wouldn't qualify under direct.
  • It scales as the veteran's service-connected condition list grows. A veteran service-connected for diabetes can claim multiple secondaries off of that one primary.

What secondary doesn't do well:

  • It requires a strong primary first. If you're appealing a denied claim for the primary, you can't easily layer a secondary on top.
  • It requires medical literature that supports the specific causal chain. Some chains are well-supported, some aren't.

When secondary is the right path:

  • Your condition developed after, and plausibly because of, an already service-connected condition.
  • There's medical literature linking your secondary condition to the primary, OR a medical examiner willing to opine on the link.

Path Four: Aggravation Service Connection

Regulatory basis: 38 CFR § 3.306. Grant rate at BVA: 41.3% across 886 cases. Use case: Pre-existing condition (documented at service entrance) was aggravated by service beyond natural progression.

Aggravation is the narrow path for conditions that already existed when the veteran entered service, but worsened during service in ways that go beyond the condition's natural course.

The classic example is bilateral pes planus (flat feet). Many recruits enter service with mild flat feet noted at entrance examination. Marching, running, and weight-bearing in combat boots can aggravate the condition during service. Under § 3.306 and the Allen v. Brown framework (covered in the aggravation deep dive), the veteran can be service-connected for the degree of aggravation.

The proof structure under § 3.306 is:

  • The condition existed prior to service (documented at entrance, typically).
  • The condition increased in severity during service.
  • The increase wasn't due to natural progression of the condition.

The third prong is the hard one. The regulations create a presumption that any worsening of a pre-existing condition during service is natural progression unless the veteran proves otherwise by "clear and unmistakable evidence." That's a heightened evidentiary standard, meaningfully harder than the "at least as likely as not" standard that governs most other VA claims.

The Allen v. Brown case (1995) is the Federal Circuit decision that framed how aggravation is measured. The compensation runs to the increment of disability attributable to the in-service aggravation, not the full disability. If the veteran entered service with 10%-severity flat feet and now has 30%-severity flat feet, the service-connected portion is the 20% increment.

Why does aggravation grant at 41.3%, roughly tied with direct?

Honestly, the 41.3% number is noisier than the other three. Only 886 cases in my dataset are classified as aggravation. The grant rate has wider confidence intervals than direct (40,026 cases) or secondary (19,189 cases) at the same level. I report the number, but I'd weight conclusions less heavily.

What I can say from the data:

  • Aggravation cases have an unusually high remand rate. The Board often can't resolve the natural-progression question on the record in front of them and sends the case back for development.
  • Aggravation works disproportionately well for a small set of conditions, bilateral pes planus is the canonical example, where the in-service aggravation is well-documented in BVA decision text and the medical literature.
  • Aggravation works poorly for conditions where the pre-service severity isn't well-documented, or where the natural progression of the condition is fast enough that "clear and unmistakable" worsening from service is hard to isolate.

What aggravation does well:

  • It captures cases that don't fit direct (because the condition pre-existed service) and don't fit secondary (because there's no service-connected primary to link to).
  • When the entrance exam documents the pre-existing condition cleanly and the service records document the aggravating activity, the path is structurally viable.

What aggravation doesn't do well:

  • The "clear and unmistakable evidence" standard is a meaningful uplift in proof burden.
  • Pre-service medical records are often thin, making the baseline severity hard to establish.

When aggravation is the right path:

  • Your entrance exam documents a pre-existing condition.
  • You have documented in-service activity or events that plausibly worsened the condition.
  • You can quantify (or have a medical examiner who can quantify) the increment of worsening attributable to service.

The 11.8-Point Spread (And Why It Matters)

I want to come back to the headline number. 46.9% presumptive grant rate, 35.1% secondary grant rate. An 11.8-point spread between the highest path and the lowest path.

That spread is not a hidden VA policy. It's not bias. It's not bad luck for secondary claimants.

It's the structure of the four paths laid out in regulation.

Presumptive removes the nexus burden entirely. Once exposure and condition criteria are met, the regulatory framework grants the connection. The hardest-to-prove element of a direct claim, the nexus opinion, is taken off the table.

Direct requires the full Shedden three-prong proof, and the third prong (nexus) is where most denials live. The 41.2% direct grant rate reflects the cases that clear all three prongs cleanly.

Secondary requires the full secondary three-prong proof under § 3.310, plus an already service-connected primary. The proof burden compounds. The 35.1% secondary grant rate reflects the cases where both layers clear.

Aggravation requires proof to a heightened standard ("clear and unmistakable evidence") that's harder than the default. The 41.3% grant rate is the cases that clear that bar.

If you mapped the four paths against their proof burdens, the grant rates would line up, and they roughly do.

The path you pick (where you have a choice) determines the regulatory framework the Board applies to your case. The framework determines what you have to prove. What you have to prove determines what evidence wins.

When You Don't Get To Choose

A lot of veterans want to know "which path should I file under?" My answer in most cases is that the path isn't a strategic choice, it's a description of the facts.

If your condition is on a presumptive list and you served in the covered location, the presumption applies. You file presumptive because the regulations apply. You don't get to opt out and file direct just because direct happens to have a marginally different grant rate.

If your condition is downstream of a service-connected primary, the secondary framework applies. You file secondary under § 3.310 because that's the regulation that governs.

If your condition pre-existed service (documented at entrance) and worsened during service, aggravation under § 3.306 is the framework.

The cases where you have a strategic choice are narrow. The most common one I see in my data is when a condition has multiple plausible theories. A veteran with PTSD who has hypertension can plausibly file:

  • Direct, if the hypertension is documented during service.
  • Secondary, if the hypertension is more reasonably attributed to chronic PTSD-related stress.
  • Presumptive, if the veteran is a Gulf War or burn-pit veteran and hypertension is on the presumptive list (it was added under PACT Act in 2022 for certain exposure categories).

Where multiple theories are plausible, the choice matters. The right answer is usually to file under the theory with the lowest proof burden, which is presumptive when available, direct when presumptive isn't, secondary when direct isn't a clean fit.

But you can also file under multiple theories simultaneously. The form doesn't make you pick one. If your VSO or attorney is doing it right, the claim narrative invokes whichever theories the evidence supports, and the RO or Board applies the framework that produces the strongest claim.

The Decision Tree

A practical order to think through your theory

If you're sitting down to figure out which connection theory applies to your case, here's the order I'd think through it.

  • Is the condition on a presumptive list, AND do you have service that meets the exposure or era criteria for that list? If yes, file presumptive. The presumption removes the nexus burden and your grant odds at the BVA are 46.9%.
  • Was the condition diagnosed during service, OR is there a documented in-service event you can link to a current condition? If yes, file direct. The Shedden three-prong framework applies. Build a nexus opinion if you don't already have one. Grant odds at the BVA are 41.2%.
  • Is the condition downstream of an already service-connected primary condition? If yes, file secondary. The § 3.310 framework applies. You need a nexus opinion linking the secondary to the primary. Grant odds at the BVA are 35.1%.
  • Did the condition pre-exist service (documented at entrance), and worsen during service beyond what natural progression would account for? If yes, file aggravation under § 3.306. You'll need to clear the "clear and unmistakable evidence" standard. Grant odds at the BVA are 41.3% but the sample is small and the remand rate is high.

Where multiple theories apply, file under all of them. The first one to clear wins. Don't make this an either-or decision unnecessarily.

If none of the four paths apply cleanly to your facts, you may not have a viable claim, or you may have a claim that's harder to package. Talk to an accredited representative if you're not sure which theory applies. The classification matters more than most veterans realize.

What I Can't Tell You About Your File

A few honest limits.

First, I cannot tell you which connection theory the RO or Board will classify your claim under. The classification is downstream of how the claim is filed, what evidence is submitted, and how the case is developed. My data shows the grant rates by classification, not by what the veteran initially intended.

Second, I cannot tell you whether your facts support a specific theory. Whether a pre-existing condition was "aggravated" beyond natural progression is a medical and evidentiary question that depends on the specific record. The same is true for whether a secondary chain is supported, whether a presumptive exposure is documented, and whether a direct in-service event is established.

Third, the connection-type grant rates in my data are descriptive of cases that reached the BVA. They are not descriptive of grant rates at the regional office, where the vast majority of claims are decided before they ever reach the Board. RO-level patterns may differ.

Fourth, my classification of connection types is parsed from BVA decision text. Cases where multiple theories were filed and one was granted while others were denied may not be cleanly classified. The 886-case aggravation sample is the noisiest in the data; the 40,026-case direct sample is the cleanest.

Fifth, the regulatory framework evolves. Presumptive lists expand. CAVC and Federal Circuit decisions reinterpret § 3.303, § 3.306, § 3.310, and § 3.307-309 over time. The grant rates I report reflect outcomes under the framework as it existed in 2023-2025. Future regulatory changes may shift the picture.

Sixth, the path with the highest grant rate isn't always the right path for your case. If you have a presumptive condition but the exposure documentation is thin, presumptive may be harder than direct. If you have a clear in-service event but no nexus opinion, direct without supporting evidence may be harder than developing a secondary theory off a stronger primary. The data is descriptive, not prescriptive.

Bottom Line

Four paths to service connection: direct (41.2% grant rate at the BVA, the default Shedden three-prong proof), secondary (35.1%, conditions caused or aggravated by service-connected primaries under § 3.310), presumptive (46.9%, the regulatory shortcut under § 3.307-309 that removes the nexus burden when exposure and condition criteria are met), and aggravation (41.3%, pre-existing conditions worsened by service beyond natural progression under § 3.306). The 11.8-point spread between presumptive and secondary is structural, not arbitrary, it reflects the different proof burdens each path imposes. Where you have a choice, file under the theory with the lowest burden. Where the facts dictate the path, file under what applies and build the strongest case for it. The right answer for most claims isn't strategic, it's a description of what your facts actually support. Know which framework applies, and you know what evidence wins.

Methodology and Limitations

  • Data source: Claim Raven's analysis of 69,078 condition records where the connection theory could be cleanly classified, from the production case_analysis dataset. Total dataset is 101,518 condition records drawn from the analyzed subset of Claim Raven's 501,000+ Board-decision library; the difference is records where the theory wasn't explicit in the decision text or where multiple theories were classified ambiguously.
  • Sample composition: direct = 40,026 cases, secondary = 19,189, presumptive = 8,975, aggravation = 886. Aggravation has the smallest sample and the widest confidence intervals.
  • Classification approach: connection theory is extracted from BVA decision text. Cases where the Board's analysis turned on a specific regulatory framework (Shedden for direct, § 3.310 for secondary, § 3.307-309 for presumptive, § 3.306 for aggravation) are classified by the dominant framework.
  • Limitation, aggravation sample: The 886-case aggravation sample is small enough that the 41.3% grant rate has meaningful noise. Direct comparisons against the larger samples should be made carefully.
  • Limitation, multi-theory filings: Cases filed under multiple theories simultaneously may be classified by the dominant theory in the Board's analysis, not by all theories filed.
  • Limitation. BVA selection bias: Connection-type classification is BVA-level. Some claims that won at the RO under one theory may have been denied under that theory and re-filed under another for BVA appeal, biasing the BVA sample toward cases where the initial theory was contested.
  • Limitation. RO-level outcomes: Grant rates reflect outcomes for cases that reached the BVA. They do not predict outcomes at the regional office level, where the vast majority of claims are decided.
  • Limitation, evolving framework: PACT Act expansions, new CAVC decisions, and changes to § 3.307-309 listings shift the picture over time. The grant rates here reflect the 2023-2025 window.
  • Limitation, individual outcomes: Aggregate grant rates by theory do not predict individual outcomes. Your specific case depends on facts no aggregate analysis captures.

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