Impacts of climate variability and climate change on regional crop yields are commonly assessed using process-based crop models. These models, however, simulate potential and water limited yields, which do not always relate to observed yields. The latter are largely influenced by crop management, which varies by farm and region. Data on specific management strategies may be obtained at the field level, but at the regional level information about the diversity in management strategies is rarely available and difficult to be considered adequately in process-based crop models. Alternatively, understanding the factors influencing management may provide helpful information to improve simulations at the regional level. In this study, we aim to identify factors at the regional level that explain differences between observed and simulated yields. Observed yield data were provided by the Farm Accountancy Data Network (FADN) and Eurostat. The Crop Growth Monitoring System (CGMS), based on the WOFOST model, was used to simulate potential and water limited maize yields in the EU15 (i.e., the old member states of the European Union). Differences between observed and simulated maize yields were analysed using regression models including: (i) climatic factors (temperature and precipitation), (ii) farm size, (iii) farm intensity, (iv) land use, (v) income and (vi) subsidies. We assumed that the highest yields observed in a region were close to the yield potential as determined by climate and considered the average regional yields as also influenced by management. Model performance was analysed with respect to spatial and temporal yield variability. Results indicate that for potential yield, the model performed unsatisfactory in southern regions, where high temperatures increased observed yields which was in contrast to model simulations. When considering management effects, we find that especially irrigation and the maize area explain much of the differences between observed and simulated yields across regions. Simulations of temporal yield variability also diverted from observed data of which about 80% could be explained by the climatic factors (35%) and farm characteristics (50%) considered in the analysis. However, effects of specific factors differed depending on the regions. Accordingly, we propose different groups of regions with factors related to management which should be considered to improve regional yield simulations with CGMS.