Intra-family crossover effects triggered by job losses have received growing attention across scientific disciplines, but existing research has reached discrepant conclusions concerning if, and if so how, parental job losses affect child mental health. Drawing on sociological models of stress and life course epidemiology, we ask if parental job losses have long-term effects on child mental health, and if these effects are conditional on the timing of, or the cumulative exposure to, job losses. We use intergenerationally linked Swedish register data combined with entropy balance and structural nested mean models for the analyses. The data allow us to track 400,000 children over 14 years and thereby test different life-course models of cross-over effects. We identify involuntary job losses using information on workplace closures, thus reducing the risk of confounding. Results show that paternal but not maternal job loss significantly increases the risk of psychotropic drug use among children, that the average effects are modest in size (less than 4% in relative terms), that they may persist for up to five years, and that they are driven by children aged 6–10 years. Moreover, cumulative exposure to multiple job losses are more harmful than zero or one job loss.