National Merit Semifinalist Cutoffs Class of 2026

Record High National Merit Scores Announced

Every year, the National Merit Scholarship Program honors approximately 17,000 students as National Merit Semifinalists based on junior year PSAT scores. Semifinalists can continue in the competition to become Finalists and, potentially, scholarship recipients. Current Semifinalists and future participants may want to read Compass’s National Merit Scholarship Program Explained for more information on the steps in the program. An additional 40,000 students are honored as Commended Students for having scores in the top 3% of all test takers. The recently confirmed cutoffs reveal that the Class of 2026 had the highest Semifinalist scores ever on the PSAT. Of the 12 largest states, 8 set new records and the other 4 tied their highest historical marks. Students in Massachusetts and New Jersey (225) would have needed to score at least a near-perfect 750 on the Reading & Writing (RW) and combine it with a 750 or 760 on Math.

The large jump points to a problem
The nearly universal increase in Selection Index cutoffs is most likely attributable to a flaw in scaling or test construction that produced higher scores on both Reading & Writing and Math. Since these sorts of scoring changes can also occur on the SAT, this post explores the implications for National Merit and college admission testing.

Scaling error best explains:

• Why there were changes across the entire score range
• Why there was a change in almost all states
• Why new records were reached in so many states, particularly the largest states

It’s the sort of shift we have seen before, but there are some added twists this time.

How cutoffs are determined
Qualifying scores (“cutoffs”) are not based on the total score for the PSAT (360-1520) but on the Selection Index, which is calculated by doubling the RW score, adding the Math score, and then dividing the sum by 10. The maximum Selection Index is 228. Students can find a historical set of cutoff data here or see how Semifinalist and Commended counts have changed state by state.

Semifinalists are allocated by state, and cutoffs are calculated by state. This means that students across the country face varying qualifying scores for Semifinalist status (the Commended level is set nationally). The cutoffs for the Class of 2026 range from 210 in New Mexico, North Dakota, West Virginia, and Wyoming to 225 in New Jersey and Massachusetts. If California is allocated 2,000 Semifinalists based on its population of high school graduates, then NMSC works down from a perfect 228 Selection Index until it gets as close as possible to that target. This year, California’s 224 included 2,172 students. A cutoff of 225 would have produced too few Semifinalists. A cutoff of 223 would have gone well over the allocation.

Below are this year’s cutoffs compared to those from prior years. The Class of 2026 figures are confirmed.

State	Class of 2026 (Actual)	Change	Class of 2025 (Actual)	Class of 2024 (Actual)	Semifinalists	Commended
Alabama	214	2	212	210	228	141
Alaska	215	1	214	209	31	24
Arizona	218	1	217	216	409	557
Arkansas	215	2	213	210	141	106
California	224	3	221	221	2172	6840
Colorado	219	1	218	216	287	579
Connecticut	223	2	221	221	193	709
Delaware	220	1	219	219	47	84
Florida	219	2	217	216	1008	1824
Georgia	220	2	218	217	620	1243
Hawaii	219	2	217	217	60	124
Idaho	215	2	213	211	90	76
Illinois	222	2	220	219	748	1888
Indiana	218	1	217	216	333	531
Iowa	214	2	212	210	138	77
Kansas	216	1	215	214	136	113
Kentucky	214	1	213	211	200	121
Louisiana	216	2	214	214	220	219
Maine	217	3	214	213	57	63
Maryland	224	2	222	221	348	1290
Massachusetts	225	2	223	222	282	1754
Michigan	220	2	218	217	470	965
Minnesota	219	2	217	216	266	438
Mississippi	213	1	212	209	153	53
Missouri	217	2	215	214	281	326
Montana	213	4	209	209	48	8
Nebraska	214	3	211	210	109	63
Nevada	214	0	214	211	185	78
New Hampshire	219	2	217	215	51	99
New Jersey	225	2	223	223	511	3199
New Mexico	210	-1	211	207	111	0
New York	223	3	220	220	992	3378
North Carolina	220	2	218	217	523	1151
North Dakota	210	0	210	207	26	0
Ohio	219	2	217	216	490	999
Oklahoma	212	1	211	208	214	39
Oregon	219	3	216	216	188	318
Pennsylvania	221	2	219	219	612	1511
Rhode Island	219	2	217	215	50	96
South Carolina	215	1	214	209	225	197
South Dakota	211	3	208	209	46	6
Tennessee	219	2	217	217	306	521
Texas	222	3	219	219	1673	4653
Utah	213	2	211	209	199	68
Vermont	216	1	215	212	27	27
Virginia	224	2	222	219	489	1912
Washington	224	2	222	220	388	1295
West Virginia	210	1	209	207	66	0
Wisconsin	215	1	214	213	287	216
Wyoming	210	1	209	207	20	0
District of Columbia	225	2	223	223	37	230
​Boarding Schools	220-225				158	652
​U.S. Territories	210	2	208	207	43	0
​​Studying Abroad	225	2	223	223	86	565
​​​Commended	210	2	208	207

 

What the PSAT tells us about the SAT
Analyzing the PSAT/NMSQT is about more than just explaining National Merit cutoffs. The PSAT also provides a unique window into the SAT program. National Merit results offer comparable year-over-year data that are more granular than what College Board provides for the SAT. The scoring anomalies we saw on the October 2024 PSAT are also likely occurring on the SAT; they’re just better disguised on the three-letter exam. Based on our historical review, scoring outliers crop up every 3 to 4 years with the PSAT. Projected across an SAT cycle, that’s potentially 2 problematic exam dates every year!

Cutoff changes
In total, 47 states saw higher cutoffs, as did the District of Columbia (225, a new record), U.S. territories and commonwealths (210), U.S. boarding schools (220-225, new records), and U.S. students abroad (225, a new record). Boarding school cutoffs are set at the highest state cutoff within the National Merit region. For students at day schools, eligibility is defined by the school’s location rather than the student’s home address.

State cutoffs always have some degree of fluctuation, especially in smaller states. The size and consistency of this year’s movements set them apart, and large states provide the best measuring stick. A 3-point increase in Maine’s cutoff might be considered unusual, but a 3-point rise in California’s cutoff demands an explanation.

The 12 largest states account for more than 10,000 Semifinalists. Their cutoffs went up an average of 2.25 points, a record change. Even the plunge in the Class of 2021, traced back to a flawed PSAT form, was more moderate.

Over the last dozen years, the majority of 2- and 3-point changes in large states’ cutoffs occurred just this year.

The bias is also seen when looking at all 50 states. The chart below shows how changes in the prior 11 years compare to the Class of 2026’s shifts. Historically, cutoffs remain unchanged approximately 30% of the time, and go up by 2 or more points only 15% of the time. This year, two-thirds of states saw increases of 2 or more points.

Was the PSAT fair? Was it wind-assisted?
In running events such as the 100m-dash, results do not qualify as world records if there is too much wind. The race results still stand; the gold, silver, and bronze medalists still finished first, second, and third. But the runners’ performances are not comparable to other races if they had a 15-mile per hour wind at their backs. While the October 2024 PSAT was likely wind-assisted, it was largely fair to those taking the test. The higher National Merit cutoffs did not alter the number of Commended Students or Semifinalists. Students were still ranked 1, 2, 3, etc.

Why the qualifier of “largely fair”?
On the digital PSAT, not all students answer the same questions. There is a pool of potential items. Nor is scaling done by a simple tally of right/wrong answers. As with the digital SAT, a specialized form of scoring called 3-parameter Item Response Theory (IRT) is used on the PSAT. IRT is a form of pattern scoring, where a student’s score is determined by which specific questions are answered correctly or incorrectly. If the parameters for questions were inaccurate and those questions only went to certain students, then the bias in scores may not have been uniform. A swirling wind could have helped some students and not others. The consistency of the upward bias, though, indicates that most students were boosted last October.

Scores provide needed insight
In the old world of paper PSATs, College Board shared select test forms with students, provided educators with performance data for questions, and released scales. None of that takes place with the digital PSAT. No items are released. No scoring parameters are provided. No performance data is shared. Students are not even told how many questions they got right or wrong. In short, visibility over the exam is available only by analyzing reported scores.

Those reported scores clearly show the upward bias. The number of students earning a 700-760 score on Reading & Writing increased from 62,000 to well over 74,000 (a 20% increase). The number of Math scores in that range went from 59,000 to approximately 78,000 (up more than 30%).

The changes at the very top were likely even more extreme. With the 223 cutoff seen in New Jersey for the Class of 2025, there were 12 score combinations that qualified a student for Semifinalist: 740RW / 750M, 740RW / 760M, etc. For the state’s 225 cutoff this year, there were only 6 combinations. It’s possible that the number of 750-760 scores went up by 50% or more.

So, the October 2024 test was easier than normal?
If easier is defined as more students able to achieve top marks, then the answer is “yes.” That doesn’t mean that the questions themselves were easier. The test’s scale is meant to adjust for differences. Somewhere along the line, things broke down.

Over the last two decades, the PSATs from 2011 (Class of 2013), 2016 (Class of 2018), 2019 (Class of 2021), and 2024 (Class of 2026) stand out as problematic. In those years, almost every state saw a change in cutoffs, and the direction and size of the change point to non-parallel forms (wind!). (The Class of 2014 also saw significant changes, but those were more of a bounce-back from the previous year.) The anomalous 2019 results could be traced back to a particularly mis-scaled form, which I wrote about at the time.

Implications for the SAT
The PSAT offers a snapshot of an entire class at a specific moment. In contrast, the SAT is administered on various dates and times, yet all results are reported as interchangeable. Some SAT takers may have wind at their backs, and some may be running directly into the wind. College Board’s goal is to prevent differing conditions or factor them out of the equation. Its objective is to ensure that the questions on each exam are nearly identical in content and difficulty (known as “parallel forms”), with any minor discrepancies accounted for through equating and scaling. However, PSAT results highlight the challenge of achieving this goal. Ultimately, some SAT administrations are going to yield higher or lower scores, just as observed with the PSAT.

Why aren’t you analyzing those SAT changes?
SAT data provided by College Board tend to obscure non-parallel results. Scores from individual test dates are not publicly shared. Even in the locked-down educator portals, scores are only reported in broad ranges. By the time College Board presents the results for a group of graduated students, the impact of non-parallel forms has been smoothed away, and College Board prefers it that way. If you can’t see scoring irregularities, did they really happen? The useful thing about the PSAT is that we can see them. National Merit cutoffs are far more granular than the 1400-1600 range that College Board reports annually for the SAT.

Non-parallel forms, norms, and student behavior
If test forms are not consistently parallel, then students have added incentive to repeat the SAT. As a test taker, why wouldn’t I want to stumble across an exam with an upward bias? The incentive is increased by the fact that superscoring locks in any upward bias and any positive error (see below) on each section of the test. Over time, the number of test dates taken by students applying to competitive colleges has increased, and testing calendars have shifted forward to allow for this. This may not be desired behavior, but it is rational behavior.

Due to upward shifts in SAT scores, traditional normative data like percentiles are insufficient for accurately measuring performance. PSAT students in the class of 2026 saw how tricky it can be comparing one’s performance to historical norms. The same problem arises on the SAT. Percentiles are provided for the three preceding class years. If there is an upward shift, it will not be fully reflected for more than three years. Unlike the ACT, College Board stopped reporting the number of students achieving each score nearly a decade ago and has never disclosed the impact of superscoring on score distribution. When assessing where an SAT score really ranks, students are not given the full picture.

In effect, College Board provides outdated track season averages for the SAT and expects them to be good enough to assess individual race results. Wind be darned.

Haven’t scores always been volatile?
Fluctuations at the individual level are different than those at the population level, although both can contribute to scoring uncertainty for students.

All tests contain inherent imprecision, known as the standard error of measurement (SEM) in psychometrics. SEM reflects that a single test can not accurately pin down a student’s “true score.” For this reason, College Board provides students with a score range, typically plus or minus 30 points, beneath their reported test scores.

Changes in the National Merit cutoffs can not be explained by SEM. Error in measurement is effectively random, and negative error and positive error cancel out when viewed over a large population. It doesn’t get much larger than the 1.5 million juniors who took the PSAT. SEM would not push scores upward.

The confidence intervals provided on student scores, however, assume parallel forms. Non-parallel forms are the likely cause of the increases on the October 2024 PSAT.

Instead of random error, scores were biased upwards, at least at the highest levels. There is strong circumstantial evidence that the October 2024 PSAT was not parallel to the October 2023 PSAT. In other words, students saw volatility (College Board’s inability to equate each test to produce equivalent scores) layered on top of typical volatility (the fluctuation of individual student scores due to SEM). The same problem arises with the SAT, it is simply hidden from view.

Fluke, shift, or trend
Was the observed bias on the PSAT a fluke, shift, or trend? The change in score distribution could be attributable to something unique to the October 2024 PSAT. We saw this happen with the paper tests in the past. There were outlier years that we might consider “flukes.”

Alternatively, we could be seeing a permanent shift upward in scores. Instead of wind at the back, are we perhaps seeing a move to a new track surface that will permanently raise scores? Equating a new test format is difficult. Equating a new format that accounts for future student behavior is even harder. Is it simply coincidence that scores jumped in both 2016 and 2024, the years after the introduction of new PSAT designs? It’s difficult to disprove a shift at its very outset.

Could the change reflect even more than a shift? Could it be a trend that will push scores higher still? This seems like the least likely possibility. Previous examples of major score differences have fallen into the fluke or shift buckets.

Other theories about the change
There are other theories as to why PSAT scores increased. For example, is the increase in PSAT scores due to better preparation? It is unlikely. I have spent much of my professional life helping students improve their test scores, so it may seem odd that I discount learning improvements or test preparation as an explanation. Practice and preparation do raise scores at the individual level. The behavior of a testing population, however, rarely changes quickly or uniformly.

The cutoffs in the largest 12 states went up either 2 points or 3 points. We should not have seen that uniformity if preparation and technique were the primary causes.

It’s Desmos’ fault
Probably not. Desmos, the powerful online calculator available for the PSAT and SAT, was available in 2023, as well. Students may have become more adept with Desmos, but that doesn’t explain why we also saw an increase in Reading & Writing scores. Further, a Desmos-linked impact should be less prominent at the highest score levels, since students capable of scoring 740-760 are less likely to see the benefit versus those scoring, say, 650-700.

Are the cutoffs explainable by a change in testing population?
The number of students taking the PSAT can change from year-to-year. The score level of those students can also change. For example, if a state begins requiring all students to take the PSAT, the average score will go down, while the number of high scorers may move up (in previous years, we saw this in Illinois and Michigan). This is a poor fit for what we saw with the PSAT. Scores went up across virtually all states. There is strong evidence that there were forces that pushed Selection Indexes up by 2 points.

Is the change attributable to the adaptive nature of the exam?
The RW and Math PSAT each have two stages. A student receives an initial set of questions. Based on their performance on that first stage, the student receives a set of easier or harder problems in stage 2. An adaptive test can more quickly narrow down a student’s score, but there is always the chance of what is known as routing error. In other words, a student with an ultimate score of 640 probably should have been routed to the harder stage 2 problems rather than the easier ones. There may be less accuracy had the student been routed to the easier set of questions. However, routing error should be neutral for the population as a whole. Further, College Board research maintains that routing error has a minimal impact on scores. Most important, students scoring at the National Merit range would have been routed to the harder stage 2 with 99+% certainty.

IRT scoring may have been a factor. Item parameters are calculated beforehand through pre-testing, where the question is included as an unscored item on earlier exams. Inaccurate parameters can lead to inaccurate scores.

The digital PSAT and SAT are shorter than their paper ancestors, and this can contribute to score instability. An individual problem or two plays a greater role on a shorter exam. While this can be offset by the adaptive nature of the test, longer is always better when it comes to test reliability. The PSAT tries to place students on a 160 to 760 scale with only 40 scored Math questions and only 50 RW questions.

Could NMSC have changed how it calculates cutoffs?
Each year, some students are unable to take the PSAT because of illness or other extenuating circumstances. These students can apply to enter the scholarship program via Alternate Entry using an SAT score. The deadline for application is generally April 1 after the PSAT, although students can use SAT scores through the June test date. In the past, NMSC has only used PSAT scores to calculate cutoffs (with an exception made during the COVID-related cancellations in 2020). Because students can take the SAT on multiple dates, their scores skew higher than PSAT scores. If NMSC were to include them in the cutoff calculations, it would likely lead to cutoff inflation. Compass has not heard that any changes were made for the Class of 2026.

Did Compass see the changes coming?
Only in part. Once PSAT scores were available in November, we noted the uptick in 1400-1520 scores and projected that the Commended cutoff would move up 2 points to 210. While upward movement was expected nationally, we did not foresee the breadth of the changes. The table below shows that there were far more high scores in the Class of 2020. The class also saw a higher Commended of 212. Yet the highest Semifinalist cutoff only reached 223. Cutoffs as high as 225 were without any precedent.

What about expectations for the Class of 2027 and beyond?
More than ever, PSAT students have to be aware that “past performance is no guarantee of future results.” In November, Compass will report on the scoring of the October 2025 exam and provide our range projections. We won’t know what future cutoffs will be, but the PSAT scores may provide clues on the question of fluke, shift, or trend.

Why does each state have its own Semifinalist cutoff if the program is NATIONAL Merit?
This is always a hot button question. NMSC allocates the approximately 17,000 Semifinalists among states based on the number of high school graduates. That way, students across the nation are represented. It also means that there are very different qualifying standards from state to state. A Massachusetts student with a 220 might miss out on being a Semifinalist. If she lived 10 miles away in New Hampshire, she would qualify.

NMSC sets a target number of Semifinalists for a state. For example, California sees about 2,000 Semifinalists every year, Michigan 500, and Wyoming 25. In each state, NMSC determines the Selection Index that comes closest to matching its target number of Semifinalists. If 1,900 California students score 222 and higher and 2,050 score 221 or higher, then the Semifinalist cutoff would be 221 (this assumes that the target is exactly 2,000). Because score levels can get crowded, it is easy for cutoffs to move up or down a point even when there is minimal change in testing behavior or performance.

No Semifinalist cutoff can be lower than the national Commended level. Cutoffs for the District of Columbia and for U.S. students studying abroad are set at the highest state cutoff (typically New Jersey). The cutoff for students in U.S. territories and possessions falls at the Commended level each year. Boarding schools are grouped by region. The cutoff for a given region is the highest state cutoff within the region.

When are National Merit Semifinalists announced for the next class?
The Commended cutoff will become unofficially known by the end of April 2026. The lists of Semifinalists will not be distributed to high schools until the end of August 2026. With the exception of homeschoolers, students do not receive direct notification. NMSC asks that schools not share the results publicly until the end of the press embargo in mid-September, but schools are allowed to notify students privately before that date. NMSC does not send Commended Student letters to high schools until mid-September. Compass will keep students updated on developments as the dates approach.

Do state and national percentiles indicate whether a student will be a National Merit Semifinalist?
No! Approximately 1% of test takers qualify as Semifinalists each year, so it is tempting to view a 99th percentile score as indicating a high enough score — especially now that College Board provides students with percentiles by state. There are any number of flaws that rule out using percentiles as a quick way of determining National Merit status.

• Percentiles are based on section scores or total score, not Selection Index
• Percentiles are rounded. There is a large difference, from a National Merit perspective, between the top 0.51% and the top 1.49%
• Percentiles reveal the percentage of students at or below a certain score, but the “at” part is important when NMSC is determining cutoffs.
• The number of Semifinalists is based on the number of high school graduates in a state, not the number of PSAT takers. Percentiles are based on PSAT takers. States have widely varying participation rates.
• Most definitive of all: Percentiles do not reflect the current year’s scores! They are based on the prior 3 years’ performance. They are set even before the test is given. And if you are going to use prior history, why not use the record of prior National Merit cutoffs rather than the highly suspect percentiles?

Entry requirements for National Merit versus qualifying for National Merit.
Your PSAT/NMSQT score report tells you whether you meet the eligibility requirements for the NMSP. In general, juniors taking the October PSAT are eligible. If you have an asterisk next to your Selection Index, it means that your answers to the entrance questions have made you ineligible. Your answers are conveniently noted on your score report. If you think there is an error, you will also find instructions on how to contact NMSC. Meeting the eligibility requirements simply means that your score will be considered. Approximately 1.4 million students enter the competition each year. Only about 55,000 students will be named as Commended Students, Semifinalists, Finalists, or Scholars. See National Merit Explained for more information.

Share this post with friends: