Retirement policies are individually designed but the majority of people of retirement age live as couples. We estimate the effects of a French pension reform on spouses' employment decisions. We use labor-force survey data, pooled over different years, on fifty thousand French couples and apply a regression discontinuity framework, also controlling for couple's unobserved heterogeneity. We conclude that the reform immediately reduced both spouses' retirement probability. The wife's retirement probability also drops by 1 to 4 percentage points if the husband is hit by the reform, and vice-versa. Instrumenting spousal retirement with legal retirement age, own retirement probability rises by 2 to 6 percentage points upon spousal retirement.