Skip to content
SVERIGEFAKTA
Schools — from European top to OECD mid-table
What happened to Sweden?

Schools

Schools — from European top to OECD mid-table

Sweden's PISA score in mathematics fell from 510 (2000) to 464 (2025). The share of pupils below baseline has doubled. PISA measures 15-year-olds in a sample of pupils.

Memory

Sweden has lost ≈ one year of learning in mathematics compared with 2003.

2000

510 p

PISA math

2025

464 p

PISA math

How to read this

Data in context

How to read this

The charts below show what the numbers say — not an interpretation. Where data point in different directions (e.g. stable trust but rising insecurity) we show both series rather than pick a narrative.

Why does 2018 stand out? The peak is a measurement artefact: Sweden excluded ~11 % of pupils from the PISA sample — twice the OECD ceiling (5 %). Mainly recent arrivals were filtered out in breach of OECD rules. The Swedish National Audit Office (RiR 2021:12) concluded the result was overstated. With a normalised sample in 2022 the average fell to 482 — in line with the underlying 2009–2015 trend.

Sveriges PISA-snitt i matematik, 2000–2025

Sveriges PISA-snitt i matematik, 2000–2025 — 2000–2025 (p)

  • 2000: 510 p; 2003: 509 p; 2006: 502 p; 2009: 494 p; 2012: 478 p — PISA-chocken; 2015: 494 p; 2018: 502 p — Sverige uteslöt ~11 % av eleverna ur urvalet — dubbelt så många som OECD:s tak (5 %). Riksrevisionen (RiR 2021:12) bedömde resultatet som överskattat.; 2022: 482 p — Återgång till mer normalt urval — närmare 2012–2015-trenden.; 2025: 464 p — Signifikant nedgång med 18 poäng mot 2022. Urvalsundersökning: punktskattning med standardfel.
200020032006200920122015201820222025
  • 2012*PISA-chocken
  • 2018*Sverige uteslöt ~11 % av eleverna ur urvalet — dubbelt så många som OECD:s tak (5 %). Riksrevisionen (RiR 2021:12) bedömde resultatet som överskattat.
  • 2022*Återgång till mer normalt urval — närmare 2012–2015-trenden.
  • 2025*Signifikant nedgång med 18 poäng mot 2022. Urvalsundersökning: punktskattning med standardfel.

The Nordics — same metric

PISA matematik — Sverige vs Norge, Danmark, Finland

2000–2022 · p · Nordics

PISA matematik — Sverige vs Norge, Danmark, Finland: Norge moves from 499 (2000) to 468 (2022) — down 31 p. Full values are available in the data table below (2000–2022, Norden).

PISA matematik — Sverige vs Norge, Danmark, Finland
PISA matematik — Sverige vs Norge, Danmark, Finland · p · 2000–2022 · Nordics
ÅrNorgeDanmarkFinlandSverige
2 000499514536510
2 009498503541494
2 018501509507502
2 022468489484482

Source:

Human consequence

Persona

A maths teacher in lower secondary school (illustrative — drawn from TIMSS teacher surveys and Skolinspektionen reports 2000–2024).

Then

  • The weakest pupils more often mastered basic algebra in year 9 (Skolverket, TIMSS Advanced 1995).
  • Instructions in Swedish could be given without heavier language adaptation for larger groups.
  • National tests were measurably comparable year over year (Skolverket evaluations).

Now

  • ≈ 27 % of pupils are below PISA baseline in maths (PISA 2022).
  • Teachers often have to explain wording before pupils can do the maths (Skolinspektionen 2022).
  • The gap between strong and weak pupils is among the widest in the OECD (PISA 2022).

2035

Sweden spends among the most per pupil in the OECD. Why does it now trail Finland and Estonia?

In PISA 2025 Sweden scores 464 p in mathematics — level with the OECD average according to Skolverket, and a significant decline from 482 p (2022). Nordic 2025 figures are not yet verified in our source chain and are therefore not shown. For 2022: Denmark higher, Finland about the same, Norway and Iceland lower. PISA is a sample survey — individual scores carry standard errors.

Sources for this page

The data comes from official primary sources.

Read the full investigation of how Sweden has changed.

Back to overview →

The question this page answers

How have Swedish pupils' knowledge outcomes evolved over time in PISA, TIMSS, PIRLS and national tests — and what can these measurements actually show?

At a glance

Key figures — the most important numbers on this page

PISA reading 2025

466

Skolverket/OECD — -21 points vs 2022 — significant decline — sample survey

PISA mathematics 2025

464

Skolverket/OECD — -18 points vs 2022 — significant decline — sample survey

PISA science 2025

485

Skolverket/OECD — no significant change — sample survey

Below baseline, reading 2025

31 %

Skolverket — share below Level 2. Was 24 % in 2022.

Below baseline, mathematics 2025

34 %

Skolverket — share below Level 2. Was 27 % in 2022.

PISA 2022 (reading/maths/science)

487 / 482 / 494

OECD — previous round, comparison point.

PIRLS grade 4 reading (2021)

521

IEA — down from 555 (2016). International mean: 500.

TIMSS grade 4 (2023)

503 / 494

IEA — mathematics/science. Down from 519/540 (2019).

What does the data show?

Objective observations — not interpretations

  • PISA 2025 (published 2026-09-08): Sweden scored 466 in reading and 464 in mathematics — significant declines versus 2022 — while science at 485 shows no significant change. All three are level with the OECD average. PISA measures 15-year-olds in a sample, not the full cohort.
  • PISA shows a clear 2018→2022 decline across all three domains — the first after the 2012–2018 recovery. OECD attributes part of the decline to the pandemic, but not all of it.
  • PIRLS 2021 was Sweden's weakest reading result since the measurement began in 2001 (521, down from 555 in 2016). TIMSS 2023 grade 4 points in the same direction.
  • The Swedish National Audit Office (RiR 2021:12) concluded that PISA 2018 was overstated: Sweden excluded ~11 % of pupils, twice the OECD ceiling. With a normalised sample in 2022 the average fell in line with the underlying 2009–2015 trend.
  • Skolverket and IFAU have documented a growing gap between average national test scores and average final grades during the 2010s — particularly in private (‘friskolor’) schools. This is grade inflation.
  • The gap between Swedish-born and foreign-born pupils is substantial in both PISA and grades. Parental education is the single strongest explanatory factor but not sufficient on its own.

Method note
PISA, TIMSS and PIRLS are sample studies with standard errors of ±3–5 points — individual year-on-year jumps are more uncertain than long trends. National tests are centrally set but locally marked. Final grades are set by teachers based on the whole school year. These four measures are not interchangeable — they measure partly different things and should be read together.

Definitions

How the numbers are counted — and what they do not cover

PISA
OECD's Programme for International Student Assessment. Measures 15-year-olds' ability to apply knowledge in reading, mathematics and science. Conducted every three years on a sample of about 6,000 Swedish pupils in 250–260 schools.PISA measures applied competency — not curriculum coverage.
TIMSS
IEA's Trends in International Mathematics and Science Study. Measures mathematics and science in grades 4 and 8 every four years. More closely aligned with curriculum content than PISA.
PIRLS
IEA's Progress in International Reading Literacy Study. Measures grade 4 reading every five years. Sweden's 2021 result was the weakest since the series began in 2001.
OECD
Organisation for Economic Co-operation and Development — 38 member states. Runs PISA and publishes country reports.
National tests
Centrally set tests in selected subjects in grades 3, 6, 9 and upper secondary. Marked locally by teachers. Designed to support consistent grading — not to compare countries.
Index points (PISA scale)
Results are expressed on a scale set so that the OECD mean equalled 500 with a standard deviation of 100 at the 2000 baseline. Roughly 40 points correspond to one school year.
Standard error
Uncertainty in the sample estimate. Sweden's PISA mean has a standard error around ±3 points — implying a 95 % confidence interval of roughly ±6 points.
Sampling and exclusion
PISA prescribes a stratified sample of 15-year-olds and allows at most 5 % exclusion (special-needs schools, short residence, etc.). Over-exclusion reduces comparability.Sweden was at ~11 % in 2018 — twice the ceiling. Source: Swedish National Audit Office RiR 2021:12.
Equivalence
In the Education Act: education shall be equivalent regardless of where it is provided. Skolverket operationalises this as between-school variation after controlling for pupil composition.
Socio-economic background (ESCS)
OECD's PISA index — a composite of parental education, occupation and material resources. Used to compare results independent of starting position.
School segregation
Uneven distribution of pupils of different backgrounds across schools. Skolverket reports both variation in composition and variation in mean outcomes.
Break in the series
Statistical term for a point where a time series can no longer be read continuously — e.g. due to a switch in test format (paper to computer in PISA 2015) or a change in sampling.
Grade inflation
Systematic rise in final grades without a matching rise in underlying performance measured by national tests or PISA. Documented by Skolverket and IFAU during the 2010s.

Primary sources

Agencies and research institutions behind this page

4

agencies/institutions

6

reports & studies

5

primary datasets

Frequently asked questions

Short answers to what is most often discussed

PISA measures 15-year-olds' ability to apply knowledge in reading, mathematics and science in everyday-style tasks. It is not a test of specific curriculum content, but of ‘reading literacy’, ‘mathematical literacy’ and ‘scientific literacy’. This is why Swedish pupils can perform differently on PISA and on a national test without either being wrong.

Related topics

Logical next steps if you want to understand the background

Recommended reading next →

Chapter 3Unread

Swedish as mother tongue

3 minRead
Chapter 3Unread

Social trust

4 minRead