We estimate the long-run personal earnings effects of winning a mayoral election, a key parameter for models of political selection. We use Italian administrative social security data from 1995 to 2017 and a sharp regression discontinuity design based on close elections. We find that, with respect to runners-up, winners experience a 16% drop in annual labour and pension earnings, that persists for ten years and fades out afterwards. Careers in politics and a two-term limit rule on mayoral elections help explain this pattern, while we find limited evidence of sectoral mobility and “revolving door” behaviour after political service. Negative effects are larger for self-employed workers – who cannot take time off from work without losing earnings – and are more persistent in Southern Italy – reflecting a labour market with higher entry costs. We also show that the compensations received by winners for their political service more than offset the negative effects on labour and pension earnings. Over a 15-year horizon, the average net present value of winning the election is equal to roughly 40,000€ – or one year of earnings for the average candidate in our sample.