Abstract We assess MPI‐ESM 1.2‐LR decadal hindcasts (lead year 1–5) to evaluate the prediction of European compound hot and dry summers with help of North Atlantic sea surface temperature (SST). Using a 64‐member ensemble and detrended data, we isolate extremes driven by internal variability. We find that initialization is crucial for capturing the 2m‐temperature‐precipitation correlation found in observational data. While the occurrence probability of hot summers correlates positively with North Atlantic SST in time, the impact of North Atlantic SST on precipitation varies. The occurrence probability of compound hot and dry summers shows no significant correlation with North Atlantic SST. Although warmer North Atlantic SST increases the predicted occurrence probability of hot summers, the Brier Skill Score indicates that these decadal predictions do not outperform climatology for rare extreme events. Utilizing North Atlantic SST shows promise for predicting hot summers, but its impact on compound summers is still unclear.