The climate in the summer months is essential for ecosystems and society. However, climate change is causing lasting changes in the characteristics of the summer climate. In order to better understand the summer climate, to capture changes in a statistically meaningful way and to develop climate sce- narios for the future, long-term climate observations and reliable climate models are needed. These three points are addressed in this thesis with the help of three main research questions. The first question examines the prevailing large-scale climate patterns which are elaborated using the climate signature of the oxygen isotope ratio in tree ring cellulose (δ18Ocel) over the past 400 years. An empirical orthogonal function analysis reveals two different modes of variability. The first mode is related to multi-seasonal anomaly patterns associated with the El Niño-Southern Oscillation. The second mode of δ18Ocel variability, which captures a north-south dipole, is associated with a regional summer atmospheric circulation pattern that has a distinct centre over the North Sea. To further exploit the climate sensitivity of δ18Ocel tree-ring records, the first grid-based reconstruc- tion of the European summer vapour pressure deficit (VPD) for the last four centuries is presented. This reconstruction is used to answer the second question of what trends in VPD have occurred in Europe over the last 400 years. The simultaneous increase in temperature and decrease in precipita- tion starts from mid-17th century in Central Europe and the Mediterranean region and relates to a positive VPD trend. This trend towards higher VPD continues throughout the observation period. In addition to studying the past summer climate with the help of a tree ring network, climate models provide valuable information on future scenarios which are highly relevant for society and ecosys- tems. Therefore, this thesis addresses the question of whether simulations with different climate models from a climate model comparison project are suitable for making reliable statements about future drought conditions and what influence the amount of greenhouse gases has on drought oc- currence. Based on a comparison between simulated and observed drought conditions for the period 1971-2000, reasonable agreement can be found between climate model simulations and the observa- tions. However, climate models cannot reproduce drought trends in observations for recent decades for large parts of the Northern Hemisphere. Furthermore, it is shown that drought occurrence is projected to increase significantly in arid regions under three different future scenarios, with the se- verity of droughts depending on greenhouse gas emissions. For regions currently less affected by prolonged droughts, such as the European continent, the climate models show that the probability of drought occurrence increases significantly under the warmest future scenario. Thus, this thesis presents new perspectives on past, present and future European summer climate using a δ18Ocel tree ring network, climate observations and climate model simulations.