How VA Combined Ratings Work: Table I and the Math Under It

Because the regulation models a whole person, and a whole person only has 100 percent of themselves to lose. Each evaluation is applied to what is left after the one before it.

38 CFR 4.25 spells the model out in its opening paragraph: 'Table I, Combined Ratings Table, results from the consideration of the efficiency of the individual as affected first by the most disabling condition, then by the less disabling condition, then by other less disabling conditions, if any, in the order of severity.'

Then it works an example on itself. 'Thus, a person having a 60 percent disability is considered 40 percent efficient. Proceeding from this 40 percent efficiency, the effect of a further 30 percent disability is to leave only 70 percent of the efficiency remaining after consideration of the first disability, or 28 percent efficiency altogether. The individual is thus 72 percent disabled, as shown in table I opposite 60 percent and under 30 percent.'

Sixty and thirty is 72. Not 90. That single line is most of what the table is doing, and the regulation prints it before it prints anything else.

Yes. They are not two methods that usually agree. They are one method printed two ways.

Every cell in Table I equals 100 minus (100 minus a) times (100 minus b), divided by 100, rounded to the nearest whole number with a half going up. Parsing the official XML of 38 CFR 4.25 and checking all 684 published cells against that expression returns zero deviations. There is no cell that departs from the formula, no special case buried in the middle of the grid, and no rule the table follows that the math does not.

One entry is missing from the body of the grid, and the regulation handles it with a note printed on the table itself: '[10 combined with 10 is 19]'. VA's own explainer repeats the same figure in plain language, saying that two disabilities each rated at 10 percent give a combined rating of 19 percent. The formula produces 19 as well. So the answer exists even where the cell does not.

So a calculator that implements the formula and a person reading the printed table should never disagree. If they do, one of them made a math error, and it is worth finding out which.

In a fixed order, two at a time, carrying the result forward. 38 CFR 4.25(a): 'To use table I, the disabilities will first be arranged in the exact order of their severity, beginning with the greatest disability and then combined with use of table I as hereinafter indicated.'

The value that comes forward is the raw table value, not a rounded one. The regulation is explicit: 'The combined value, exactly as found in table I, will be combined with the degree of the third disability (in order of severity).'

Here is the regulation's own three-evaluation example, verbatim: 'Thus, if there are three disabilities ratable at 60 percent, 40 percent, and 20 percent, respectively, the combined value for the first two will be found opposite 60 and under 40 and is 76 percent. This 76 will be combined with 20 and the combined value for the three is 81 percent. This combined value will be converted to the nearest degree divisible by 10 which is 80 percent.'

Now a case that lands on a boundary, so the last step matters. Take evaluations of 60 percent, 30 percent and 10 percent. Order of severity is already 60, 30, 10.

Step one: 60 with 30. Table I opposite 60 and under 30 reads 72, which is the example the regulation prints itself.

Step two: 72 with 10. The exact value is 100 minus 28 times 90 divided by 100, which is 74.8. Table I carries it as 75.

Step three: convert, once. A combined value of 75 ends in 5, so it is adjusted upward. The final degree of disability is 80.

Three evaluations that add to 100 on a phone calculator produce a combined value of 75 and a final figure of 80. Both of those numbers are correct at the same time. They are answers to different questions.

There are three separate roundings in this procedure, and only one of them lands on a multiple of ten. Confusing them is the single most common way an otherwise careful reader arrives at a wrong number.

The first happens at every pairwise step. Table I cells are whole numbers, so reading a value off the table rounds the exact product to the nearest whole number. Nothing about that rounding involves a ten.

The second happens only when the bilateral factor is in play. Ten percent of a combined value is usually a fraction, and it is carried forward as a whole number. That step is covered on the bilateral page rather than here, because it runs before this procedure starts.

The third is the conversion to a multiple of ten, and 38 CFR 4.25(b) states in terms how often it happens: 'The conversion to the nearest degree divisible by 10 will be done only once per rating decision, will follow the combining of all disabilities, and will be the last procedure in determining the combined degree of disability.'

Only once. After everything has been combined. Last.

Because the regulation says so, and it says so twice, once as a rule and once as a worked case.

The rule, at 38 CFR 4.25(a): 'This combined value will then be converted to the nearest number divisible by 10, and combined values ending in 5 will be adjusted upward.'

The case, in the next breath: 'Thus, with a 50 percent disability and a 30 percent disability, the combined value will be found to be 65 percent, but the 65 percent must be converted to 70 percent to represent the final degree of disability.' A 65 does not go to the nearer neighbor of a tie. It goes up.

Ordinary rounding runs the other way, downward on a 4. The same paragraph shows that too: 'Similarly, with a disability of 40 percent, and another disability of 20 percent, the combined value is found to be 52 percent, but the 52 percent must be converted to the nearest degree divisible by 10, which is 50 percent.'

Which puts an unusual amount of weight on a single point of combined value. A combined value of 44 converts to 40. A combined value of 45 converts to 50. Those two are one point apart and ten points apart at the same time, and there is nothing gradual in between them.

For a reader checking their own decision letter, that has a practical consequence. The percentage printed on the letter is the converted figure, and the pre-conversion combined value is often not printed at all. Checking the math means recomputing from the individual evaluations the letter lists, keeping every intermediate value exactly as Table I gives it, and converting only at the end. If the recomputed figure and the letter disagree, the useful next step is a free accredited VSO representative, who will read the letter with a veteran and does not charge for it. Accredited attorneys and agents also exist and may charge.

You get a wrong answer, and not always in the same direction. This is the classic bad implementation, and it survives because the first rounding it performs is one the regulation genuinely does print.

Run 60, 30 and 10 the wrong way. Sixty and thirty is 72, and 72 rounds to 70. Seventy with 10 is 100 minus 30 times 90 divided by 100, which is 73, and 73 converts to 70. Final answer 70. Done correctly, as worked above, the same three evaluations give 75 and convert to 80. Ten points lost to one premature rounding.

Now run 50, 30 and 20 the wrong way, and watch the error flip. Fifty and thirty is 65, which the regulation itself converts to 70 in its own example. Rounding there and carrying 70 forward: 70 with 20 is 76, which converts to 80. Done correctly, the 65 is carried as 65, and 65 with 20 is 100 minus 35 times 80 divided by 100, which is 72. That converts to 70.

Same evaluations, ten points apart, and the wrong method overshot in the second case after undershooting in the first. There is no safe direction to be wrong in. That is why 4.25(b) puts the conversion last and allows it once per rating decision. It is the sentence that makes the table and the formula agree.

This is math about the table. It holds for any set of evaluations fed into it, because each one acts on what is left, and less is left every time.

Take a 10 percent evaluation and see what it does to combined values of different sizes. Added to nothing, it produces 10. Added to a combined value of 50, it produces 55, a gain of five points. Added to 80, it produces 82. Added to 90, it produces 91.

The pattern is exact: combining 10 with a value V adds (100 minus V) divided by 10 points before rounding. At V of 90, that is one point. The same math is why a combined value can climb toward 100 without ever quite reaching it by combination alone, and why the last stretch is the slowest part of the table.

Two things, and both of them change the inputs rather than the method.

The first is separate rating of conditions that came from one source. 38 CFR 4.25(b): 'Except as otherwise provided in this schedule, the disabilities arising from a single disease entity, e.g., arthritis, multiple sclerosis, cerebrovascular accident, etc., are to be rated separately as are all other disabiling conditions, if any. All disabilities are then to be combined as described in paragraph (a) of this section.'

The second is the bilateral factor at 38 CFR 4.26, and it runs first. Where a partial disability affects both arms, both legs, or paired skeletal muscles, the two sides are combined and then 10 percent of that value is added, expressly not combined, and the regulation says this happens 'before proceeding with further combinations, or converting to degree of disability.' The result is then treated as one disability for ordering and for every combination after it.

That is the whole of what 4.26 does at this page's altitude. It has its own math, its own definition of a pair, and a 2023 exception that changes the answer in some cases. All of that is on the bilateral factor page.

Worth being precise about one thing here: 38 CFR 4.25 contains no occurrence of the word bilateral and no cross-reference to 4.26. The combined ratings table knows nothing about pairs. It receives a number that the bilateral rule has already produced.

It does not decide the individual evaluations. Those come from the rating criteria in 38 CFR part 4 applied to what the medical record documents about function, condition by condition. Table I takes those percentages as given. Change what the record says and the inputs change; the table does not.

It does not decide payment. The monthly amounts sit in VA's rate tables, currently the ones effective December 1, 2025, and they depend on the combined degree of disability along with dependents.

And it is not the only route to a total evaluation. 38 CFR 4.16 carries total disability based on individual unemployability, which turns on the ability to follow a substantially gainful occupation rather than on the combined value the table returns.

Whether a particular set of evaluations was combined correctly on a particular decision letter is claims work. A VA-accredited VSO representative will do it with a veteran at no cost.