ABSTRACT The sensitivity to the observational reference has been reported in recent studies, highlighting the importance of observational uncertainty in climate research. These studies stress the importance of properly comparing available datasets, recognising their respective strengths and limitations. Here, we conduct a comprehensive analysis of the various datasets of maximum and minimum daily temperatures available for mainland Spain. We examined 10 publicly available daily gridded datasets of maximum and minimum temperatures, analysing multiple evaluation dimensions to identify the key strengths and limitations of each dataset: statistical distribution, extreme events, temporal structure and spells and spatial patterns. We conclude that observational uncertainty is greater for minimum temperatures than for maximum temperatures. This uncertainty is not strictly linked to the type of dataset (interpolation, analysis or reanalysis) or its spatial domain (national, European or global) but rather to specific datasets which vary depending on the analysis dimension. Overall, the most stable dataset across all evaluated indices is STEAD, whereas the PTI‐Clima v0 dataset exhibits some underestimation of extremes and spells but performs well in capturing central parameters and temporal correlations.