AbstractWe study the effect of Law 81/1993, which introduced a different rule for the election of mayors, on the technical efficiency of Italian cities over the period 1998–2006. Since 1993, municipalities below 15,000 inhabitants vote with a single‐ballot system, whereas cities above 15,000 inhabitants are subject to a double ballot. We first estimate the output‐oriented technical efficiency of municipalities through data envelopment analysis, and then we perform a regression discontinuity design analysis by exploiting the exogenous change in the electoral scheme at the 15,000 inhabitants cut‐off. We find evidence that municipalities voting under a double‐ballot rule show lower levels of efficiency—in terms of provision of public goods—compared with cities voting under a single‐ballot scheme, by about 6.5%. This effect is driven by inefficiencies registered in specific sectors, that is, police, education, and maintenance of roads. Furthermore, the overall lower efficiency of municipalities voting under runoff is related to the different level of educational attainment of elected mayors, whereas it is not driven by the features of Law 81/1993 relating to the number of lists forming a coalition in support of mayoral candidates and to disjoint votes.