AIMS: Previous studies found increased cardiovascular mortality during hot days, while emergency hospital admissions were decreasing. We explored potential underlying reasons by analysing clinically similar cardiovascular disease groups taking into account primary, underlying and immediate causes of death. METHODS AND RESULTS: We assessed associations of daytime maximum temperature in relation to cardiovascular deaths and emergency hospital admissions between 1998 and 2016 in Switzerland. We applied conditional quasi-Poisson models with non-linear distributed lag functions to estimate relative risks (RRs) of daily cardiovascular mortality and morbidity for temperature increases from the median (22°C) to the 98th percentile (32°C) of the warm season temperature distribution with 10 days of lag. Cardiovascular mortality (n = 163,856) increased for total cardiovascular disease (RR 1.13, 95% confidence interval [CI] 1.08-1.19) and the disease groups hypertension (1.18, 1.02-1.38), arrhythmia (1.29, 1.08-1.55), heart failure (1.22, 1.05-1.43) and stroke of unknown origin (1.20, 1.02-1.4). In contrast, emergency hospital admissions (n = 447,577) decreased for total cardiovascular disease (0.91, 0.88-0.94), hypertension (0.72, 0.64-0.81), heart failure (0.83, 0.76-0.9) and myocardial infarction (0.88, 0.82-0.95). Opposing heat effects were most pronounced for disease groups associated with diuretic and antihypertensive drug use, with the age group ≥75 years at highest risk. CONCLUSIONS: Volume depletion and vasodilation from heat stress plausibly explain the risk reduction of heat-related emergency hospital admissions for hypertension and heart failure. Since primary cause of death mostly refers to the underlying chronic disease, the seemingly paradoxical heat-related mortality increase can plausibly be explained by an exacerbation of heat effects by antihypertensive and diuretic drugs. Clinical guidelines should consider recommending strict therapy monitoring of such medication during heatwaves, particularly in the elderly.