Early leavers from education and training
- Eurostat dataset
- sdg_04_10
- Reference period
- 2025
- Unit of measure
- % of population aged 18-24
Key estimates
- EU aggregate (2025): 9.10 % of population aged 18-24.
- Highest value: Türkiye at 17.9 % of population aged 18-24; lowest: Croatia at 2.10 % of population aged 18-24.
- Cross-country mean 8.12 % of population aged 18-24, standard deviation 3.81, coefficient of variation 46.9% across 33 reporting countries.
- Aggregate change 2014→2025: -24.5% on a constant panel of 32 countries.
How many young Europeans leave education too early? Eurostat dataset sdg_04_10 reports early leavers from education and training for 33 reporting countries, with 2025 as the most recent period carrying broad coverage. The European aggregate stands at 9.10 % of population aged 18-24. Early school leaving is the strongest single predictor of lifetime labour market disadvantage.
The five highest values in 2025 are recorded by Türkiye (17.9), Iceland (16.7), Romania (15.5), Germany (13.1), Spain (12.8), all expressed in % of population aged 18-24. At the opposite end of the distribution sit Croatia (2.10), Greece (3.00), Bosnia and Herzegovina (3.00), Ireland (3.60), Norway (3.80). The ratio between the highest and the lowest observation is 8.52, while the median country reports 7.40 % of population aged 18-24 against a mean of 8.12 % of population aged 18-24.
Dispersion is wide: the standard deviation across countries is 3.81 % of population aged 18-24, giving a coefficient of variation of 46.9%. The mean sits above the median, which indicates that the distribution is pulled by the upper tail rather than being symmetric. Any European average quoted for this indicator therefore describes a synthetic country that few Member States resemble.
Between 2014 and 2025 the summed value across the 32 countries reporting in both periods moved by -24.5%. The largest relative increases are observed in Cyprus (+44.1%), Austria (+42.9%), Germany (+37.9%); the largest decreases, or the smallest increases, in Norway (-67.5%), Greece (-66.7%), Portugal (-64.7%). Because the panel is held constant, this change is not an artefact of countries entering or leaving the sample.
Lower values are the policy-preferred direction, so the top of the distribution identifies where the burden concentrates: bringing the highest observation down to the median would remove 10.5 % of population aged 18-24 from the European total for that country alone. The estimates above are reproducible: the dataset identifier, filter dimensions and reference period are stated in the methodological note, and the series can be re-downloaded from the Eurostat dissemination API at any time.
Methodological note
Source: Eurostat, dataset sdg_04_10 ('Early leavers from education and training by sex'), extracted from the Eurostat dissemination API (JSON-stat 2.0) on the dataset update of 2026-06-11. Filter dimensions: sex=T; unit=PC. Unit of measure: % of population aged 18-24. Reference period: 2025. Geographic perimeter: national reporting units with two-character geo codes (33 countries with a non-missing observation); European and euro-area aggregates are excluded from the cross-country statistics and reported separately. Descriptive statistics (mean, median, population standard deviation, coefficient of variation) are computed unweighted over reporting countries. Change over time is computed on a constant panel: only countries with a non-missing observation in both the base and the reference period enter the calculation, which removes composition effects but may differ from the officially published aggregate. No imputation, seasonal adjustment or re-scaling has been applied beyond what Eurostat performs at source. Flagged provisional and estimated observations are retained as published.
Source data
The underlying series can be inspected and re-downloaded from the Eurostat data browser.