We evaluate the importance of high-resolution proxies for boreal winter (October to March) and summer (April to September) European and North Atlantic temperature reconstructions. Multiple regression, backward elimination and cross-validation techniques are used to achieve this goal. The analysis considers natural proxies and synthetic pseudo-documentary indices. The results suggest that the most valuable predictors for European winter temperature are documentary-based indices, while tree-rings performed best for the warm season. It was also shown that the temperature signal in a speleothem from Scotland may be used for further winter and summer temperature reconstructions over parts of the Atlantic Ocean. This study represents a step towards the optimal selection of proxies which will improve temperature reconstructions.