Extreme climate events, such as droughts and heatwaves, can have large impacts on the environment. Disentangling their individual and combined effects is a difficult task, due to the challenges associated with generating controlled environments to study differences in their impacts. One approach to this problem is creating artificial climate forcing with varying magnitude of univariate and compound extremes, which can be applied to process-based impact models. Here, we propose and describe a set of six 100-year long climate scenarios with varying drought-heat signatures that are derived from climate model simulations whose mean climate is comparable to present-day climate conditions. The changes in extremes are most notable in the 3 months in which vegetation activity is highest and where arguably hot and dry extremes may have the largest impacts. Besides a control scenario representing natural variability (Control), one scenario has neither heat nor drought extremes (Noextremes), one has univariate extremes but no compound extremes (Nocompound), one has only heat extremes but few droughts (Hot), one has only droughts but few heatwaves (Dry), and one has many compound heat and drought extremes (Hotdry). These scenarios differ only moderately in their global mean climate (about 0.3°C in temperature and 6% in precipitation) and do not contain any long-term trends. The data are provided on a daily timescale over land (except Antarctica and parts of Greenland) on a regular 1° × 1° grid. These scenarios were constructed primarily to investigate the impact of varying drought-heat signatures on vegetation and the terrestrial carbon cycle. However, we believe that they may also prove useful to study the differential impacts of droughts and heatwaves in other areas, such as the occurrence of wildfires or crop failure. The data described here can be found on zenodo (https://doi.org/10.5281/zenodo.4385445, Tschumi et al., 2020).