2026 CAASPP Results: What Changed Across California?

Elsewhere on this page each rate is rounded to a whole percent first, so a change always matches the two rates beside it. These are the source’s two-decimal rates, spring 2025 to spring 2026:

English and math improved more slowly than the year before, and science faster. The 2025 CDE release reported gains of 1.8 points in English and math and 2.0 in science; the 2026 release reports a smaller step in English and math and a larger one in science.

A higher percentage means a larger share of the 2026 results reached the top two levels. The state’s release credits investments in tutoring and other supports, but yearly totals like these can’t isolate what any program did.

The four levels run from Minimal through the Developing level and Proficient to Advanced. Those names replaced Standard Not Met through Standard Exceeded starting with 2025–26 reports, and the state’s April 2026 note for families says the test questions, format and scoring stayed the same. Our score guide explains what each level means. Every grade below uses the top-two share.

Elementary grades: math edged up, English stayed flat

Third-, fourth- and fifth-grade English were flat, at 44%, 46% and 49%. Math rose by a point or less in each of the three grades.

If your child is in grades 3–5, the statewide English story is flat, so your own school’s grade in the lookup above matters more than the headline. The state totals can’t say which reading or writing skills explain it.

Middle grades: seventh and eighth grade led

Sixth-grade English was flat. Seventh and eighth each rose one point, to 51% and 49%. In math, sixth grade was flat, and seventh and eighth each rose two points, the largest gains of any grade in English or math.

Even after that gain, eighth-grade math has the lowest top-two share of the three middle grades, at 36%. Each grade’s test covers that grade’s own material, so the percentages don’t measure the same body of knowledge.

High school: one tested grade

Eleventh-grade English held at 57%. Math rose one point, to 32%. That is one tested grade, not every high-school student.

SchoolScope’s high-school Scope Scores and percentiles still use spring 2025 tests; these 2026 results appear beside them, separately. Elementary and middle-school Scope Scores and percentiles use spring 2026 tests. Check the year beside each measure.

In 2026, 51% of science results were in the Developing level, 13% at Minimal and 36% at Proficient or Advanced. A top-two percentage puts results in the Developing and Minimal levels on the same side of the line, so it can’t tell you how many children were close.

The drop and the rebound differed by grade. Third-grade English fell seven points from 2019 to 2022 and has regained two. Eighth-grade English fell two and regained two. Eighth-grade math had both the biggest drop and the biggest rebound: eight points down and seven back.

Eighth-grade English had 49% in the top two levels in both 2019 and 2026. But in 2026, a larger share was at Minimal and a larger share at Advanced, while fewer were at Proficient. Its Minimal share did fall from 2025 to 2026, from 28% to 27%, and it is still above 2019’s 26%.

The chart below shows all four levels for every tested grade since 2015. Choose a grade to see whether both ends moved, or only one.

Here, “recovery” means a grade’s reported rate returning toward its 2019 level. It can’t tell us how much learning individual children lost or regained. Rates had been rising before the disruption, from 44% to 51% in English and from about a third to 40% in math between 2015 and 2019. Getting back to 2019 would recover that level. It wouldn’t show where results might have reached without the disruption.

2019 is the last year before the testing disruption. Testing was waived in 2020, and 2021 participation was uneven, as CDE’s January 2022 release explains. We leave 2021 out of this comparison. A rise from 2022 to 2026 does not mean results rose every year in between.

CDE’s English and math score-range tables keep the same achievement-level boundaries. Some lowest and highest possible scores changed in 2021. The 2025 assessment FAQ says results from the shorter tests can be compared with the earlier full tests. Students and participation can change each year.

We also compared a fixed set of schools while keeping each school’s weight the same. Younger grades still had larger remaining gaps. That check holds the included campuses and weights fixed, but their students can change. It excludes schools without comparable records.

We grouped eligible schools by their 2019 results, separately for each grade and subject. In English grades 3–5, schools in the lowest-scoring fifth generally rebounded more from 2022 to 2026 than those in the highest-scoring fifth. These are historical groups, not current school-quality rankings.

In third-grade English, the typical school’s change was +4 points in the lower group and −2 points in the upper group. Their typical 2026 rates remained far apart: 24% and 72%. In eighth-grade math, both groups rebounded by five points.

“Typical change” is the median of schools’ individual changes. “Typical rate” is the median rate in one year. Subtracting two group medians can give a different answer, even a different direction.

This comparison includes active California public and charter schools in our regular-school comparison class, observed in 2026. Each school-grade-subject result needs a published rate and at least 30 scored results in 2019, 2022, 2025 and 2026. Schools form five roughly equal groups, keeping tied 2019 rates together. Checks with higher minimum counts, complete annual records and different weights retained the elementary-English pattern. None identifies a teaching effect or holds the student population fixed.

The difference between White and Black students’ math shares narrowed by two points, to 29 points. English dipped for Asian students, by one point, and was flat for White and Filipino students. No race or ethnicity group in the chart below fell in math or science.

The rate for students CDE counts as socioeconomically disadvantaged rose in every tested grade and subject, including two points in eighth-grade math. Their math difference with other students narrowed by one point, to 31 points. CDE’s category is broader than household income: it also includes parent education and several eligibility or student-status criteria.

For English learners, English was flat at 10%, while math and science each rose about a point. Students who become fluent leave this group, so its change can’t show how the same learners progressed. CAASPP English tests English language arts; the separate ELPAC tests English-language proficiency.

Rates for students with disabilities were flat in English and math and up about a point in science. These results don’t include the separate California Alternate Assessments, so they don’t cover every tested student with a disability. Rates for students experiencing homelessness held flat in English and rose in math and science. Rates for foster youth were flat in English and math.

A student can be in several of these groups. The chart keeps each group’s current rate beside its change, with exact rates and scored counts in its table.

At the source’s precision, the fixed-mix check gives math 1.0 point against the headline 1.1, English 0.4 against 0.5, and science 3.2 against 3.3.

The share of students counted as socioeconomically disadvantaged dipped slightly, from 63.6% to 63.1%, and that group also made up a smaller share of test results. Holding each group’s share fixed holds the broad mix steady, not the same children. Students within each group can still change, so it can’t show how much anyone learned.

Fewer tests do explain one English oddity. English had 2,931,356 scored results in 2025 and 2,877,492 in 2026. So its share rose even though 13,258 fewer results reached the top two levels. In math and science, the number of top-two results rose, by 5,611 and 33,397. Each subject is a separate count; adding them would count some children more than once.

The enrollment files are fall Census Day snapshots across all grades, separate from the tested-grade populations in the spring results. Transitional kindergarten, or TK, grew by 35,743 students to 213,313. Subtracting TK from each year’s total leaves a decline of 110,704 students, almost 2%. Kindergarten enrollment fell by 9,611.

The number classified as socioeconomically disadvantaged fell by 73,499, to 3,618,062. English-learner enrollment fell by 83,159, while enrollment of students with disabilities rose by 23,717. These categories overlap. The totals do not tell us how many children reclassified, moved away or changed disability status.

Enrollment fell in 45 of California’s 58 counties. Los Angeles County accounted for 32,953 of the statewide net loss of 74,961, 44%. Eight of the 13 counties with higher enrollment had fewer students after TK was subtracted. These are countywide school enrollments, not counts of all children living there.

Charter enrollment fell by 2,077, or 0.3%; noncharter enrollment fell by 72,884, or 1.4%. The charter share nevertheless rose from 12.5% to 12.7%, because noncharter enrollment fell faster.

Demographic headcounts and shares can tell different stories too. Hispanic or Latino enrollment fell by 48,064, while its share barely changed, from 56.1% to 56.0%. White enrollment fell by 31,076. Asian and multiracial counts rose by 1,853 and 3,574. The not-reported race category also grew, so these figures cannot be read as a record of individual students changing categories.

CDE’s official summary reports 9,947 schools and 1,011 districts. Our enrollment-file count is slightly different: 9,946 enrolled school records, compared with 9,982 in the prior file. Missing enrollment records do not establish school closures. The source notes below explain which records we counted.

Among 9,874 school codes present in both enrollment files, 5,849 lost students, 3,827 gained and 198 were unchanged. A continuing code can still change its grades or school type. For local planning, the school’s own enrollment by grade is more useful than treating the statewide decline as uniform.

Within the economic groups, the disadvantaged group’s math top-two count rose by 9,635, while the count outside the group fell by 4,024. Together those changes produce the statewide increase of 5,611; they do not track individual children crossing a threshold.

A grade’s 2025 and 2026 results usually come from different classes of children, so a drop isn’t a verdict on one year’s teacher. If your school’s grade fell, ask how it fits what the school sees in classwork and local assessments.

Bring your child’s report to the teacher and ask how it fits the reading, writing or math your child is doing now.

The shares on this page show how many results reached each level. CDE also measures matched-student growth, which compares each student’s score with what students who scored similarly the year before went on to score. The 2026 release schedule puts that growth data in its December supplemental release.

Data source: California Department of Education. The latest-year analysis uses original statewide Smarter Balanced English/math and CAST science research rows for spring 2025 and spring 2026, selected by state identifier, test, student group and grade. Each percentage uses Total Students Tested with Scores. Group comparisons use each subject’s native All grades rollup, rather than an average of grade or school percentages. Science’s High school row is CDE’s native aggregate of grades 10–12. Alternate assessments are excluded.

The overall science comparison is 32.65% in 2025 and 35.99% in 2026, based on 1,298,683 and 1,270,968 scored results. English uses 2,931,356 and 2,877,492; math uses 2,946,081 and 2,880,024. Each subject has its own population. A child tested in more than one subject contributes a separate result to each; these counts must not be added and described as unique students.

We use published combined percentages. Changes and differences between groups are SchoolScope’s arithmetic. The new comparison visual includes all eight named race/ethnicity groups, the published male/female groups, economic and disability comparisons, English learners excluding reclassified students, homeless students and foster youth. These are selected reporting categories, not every available subgroup or cross-tabulation. The latest-results comparison chart’s full table includes published rates and scored counts and works without JavaScript. We have not tested the statistical significance of the reported differences.

The enrollment section uses CDE’s Census Day files, selected at the statewide or county level with Charter ALL and the stated reporting category. These are first-Wednesday-in-October snapshots; districts and independently reporting charters with uncertified submissions are excluded. School-record counts use unique county/district/school codes and exclude school codes 0000000/0000001. Missing records do not establish closures. The latest raw file has 10,558 school-level all-student rows; the two excluded codes account for 612 rows and 14,562 students.

The economic-mix calculation uses exact top-two achievement counts divided by scored results. For each subject, it applies the 2025 disadvantaged/non-disadvantaged scored shares to the two groups’ 2026 rates, then compares that result with 2025. These two groups exhaust that subject’s scored population. Count-derived changes can differ slightly from subtracting rounded published percentages: English’s exact increase is 0.453 points, while the rounded percentages give 0.46. Holding broad group shares constant does not hold individual children or all other characteristics constant. CDE’s accountability definition of economic disadvantage includes criteria beyond income; it does not establish identical historical implementation in every assessment release.

The school lookup matches school identifier, grade and subject across 2025 and 2026. Each subject needs a published rate and at least 30 scored results in both years. A school can have a comparison for one subject without the other. Direction counts require both subjects and include active public and charter schools seen in the 2026 release. Each school counts once within a grade, regardless of size; it can appear in several grades. Those counts describe eligible schools, while statewide rates describe scored results. If a comparison is unavailable, open the school’s profile for its available evidence.

The recovery charts use all-student English and math rows for 2019, 2022 and 2026; the all-levels chart uses every year since 2015 except 2020 and 2021. We did not find a strong, general lower-versus-upper school-group difference in the latest year alone. Grouping schools by the starting or ending year of a change could reverse the pattern, so the separate 2022–2026 school-group analysis uses 2019 groups.