Purpose: This study examines the returns from education for three distinct groups: always employees, dabblers in self-employment and sustained self-employed individuals. We aim to understand how educational attainment translates into earnings across these employment types in the UK. Design/methodology/approach: Using data from the British Household Panel Survey (BHPS) and Understanding Society: The UK Household Longitudinal Study (UKHLS), we employ a random effects (RE) model to account for unobserved individual characteristics and the Heckman selection model to address self-selection bias, ensuring accurate estimates of educational returns. Findings: Our findings indicate that sustained self-employed individuals benefit more from education compared to dabblers and, in certain cases, traditional employees. Dabblers with postgraduate education report higher returns than always employees, but those with lower educational levels experience disadvantages due to their intermittent labour market engagement. Originality/value: This study introduces new evidence on the heterogeneity of educational returns for self-employed individuals in the UK, providing a novel comparative analysis of different employment types and highlighting the unique challenges and outcomes related to educational attainment and earnings.