Successful weather forecasts on subseasonal time-scales can support societal preparedness and mitigate the impacts of extreme events. Heatwaves in particular can, in certain cases, be predicted on time-scales of several weeks in advance. Heatwave predictability is commonly assessed in terms of heatwave intensity. In addition to heatwave intensity, we assess the predictability of heatwave onset and duration, which are crucial components of early-warning systems and emergency preparedness plans. The forecast skill of heatwaves is investigated over the European region in the subseasonal forecasting system of the European Centre for Medium-Range Weather Forecasts (ECMWF). The heatwaves are first detected in ERA-Interim reanalysis data over the period 1998–2017 and then allocated into six clusters in the following regions: Black Sea (BSea), Russia (Ru), Western Europe (WEu), North Sea (NSea), Scandinavia (Sc), and Eastern Europe (EEu). The European regions with the highest predictability in heatwave onset and duration are the clusters Ru, Sc, and NSea. The WEu cluster has the lowest bias in heatwave intensity and is found to be the most predictable region in terms of the number of heatwave events with predictable intensity at lead week 2. Heatwave intensity is generally found to be the most predictable characteristic of European heatwaves, being predictable by the model ensemble mean up to lead times of 3 weeks. Furthermore, this analysis identifies the most predictable heatwaves, allowing for a further investigation of the physical mechanisms and heatwave characteristics leading to enhanced heatwave forecast skill over different European regions.