Objective We investigated how educational performance during secondary education evolves in the years before, during, and after parental divorce. Background Theoretical models predict different divorce effect patterns in the school year of the divorce and in the years around it. However, we are not aware of any existing studies attempting to measure this pattern. Method The study was based on unique panel data capturing parental divorce and school results for 2,685 Belgian pupils. We employed this panel to correct for time‐invariant, unobservable drivers of educational performance, as well as to take into account time‐variant unobserved drivers. Results The results indicate that the divorce penalty peaks in the school year of the divorce. A zero treatment effect in the preceding and subsequent years cannot be rejected. Moreover, simpler econometric strategies tended to overestimate the divorce penalty. Conclusion Our findings are mostly in line with the parental adjustment and life stress theories. From a policy perspective, we suggest that it is essential to support children in the months before and after parental divorce, while special treatment in later years may be less necessary.