Abstract. Extreme weather conditions like heatwaves and drought can substantially affect tree physiology and the emissions of isoprene. To date, however, there is only limited understanding of isoprene emission patterns during prolonged heat stress and next to no data on emission patterns during coupled heat–drought stress or during post-stress recovery. We studied gas exchange and isoprene emissions of black locust trees under episodic heat stress and in combination with drought. Heatwaves were simulated in a controlled greenhouse facility by exposing trees to outside temperatures +10 °C, and trees in the heat–drought treatment were supplied with half of the irrigation water given to heat and control trees. Leaf gas exchange of isoprene, CO2 and H2O was quantified using self-constructed, automatically operating chambers, which were permanently installed on leaves (n = 3 per treatment). Heat and combined heat–drought stress resulted in a sharp decline of net photosynthesis (Anet) and stomatal conductance. Simultaneously, isoprene emissions increased 6- to 8-fold in the heat and heat–drought treatment, which resulted in a carbon loss that was equivalent to 12 and 20 % of assimilated carbon at the time of measurement. Once temperature stress was released at the end of two 15-day-long heatwaves, stomatal conductance remained reduced, while isoprene emissions and Anet recovered quickly to values of the control trees. Further, we found that isoprene emissions covaried with Anet during nonstress conditions, while during the heatwaves, isoprene emissions were not related to Anet but to light and temperature. Under standard air temperature and light conditions (here 30 °C and photosynthetically active radiation of 500 µmol m−2 s−1), isoprene emissions of the heat trees were by 45 % and the heat–drought trees were by 27 % lower than in control trees. Moreover, temperature response curves showed that not only the isoprene emission factor changed during both heat and heat–drought stress, but also the shape of the response. Because introducing a simple treatment-specific correction factor could not reproduce stress-induced isoprene emissions, different parameterizations of light and temperature functions are needed to describe tree isoprene emissions under heat and combined heat–drought stress. In order to increase the accuracy of predictions of isoprene emissions in response to climate extremes, such individual stress parameterizations should be introduced to current BVOC models.