We study the effect of solar variability on temperatures recorded in three European stations with the longest gapless series available (Prague, Bologna and Uccle). Following a pattern recognition approach, we partition daily temperature “indices” (minimum, maximum and range) in two separate classes with respect to the level of solar activity (high H vs low L 11 year cycles). Using the two-sample Kolmogorov-Smirnov statistics, multiple shuffles of data, and other partitions, we demonstrate that the separation between the probability distribution functions of H and L temperatures is statistically significant and robust. We find that average annual variations for the H and L classes display common and site-specific patterns. For practically all series considered, differences between graphs of annual change for the two classes H and L are large (∼1 °C). Solar activity accounts, at least in part, for the multi-decadal variations in temperature observed at these European sites in the past two centuries.