The impact of high temperatures on mortality is well known, but not all deaths that occur during heat waves can be explained by this effect. We evaluated whether an additional mechanism caused by periods of sustained heat without nightly cooling influenced mortality during the European heat wave in 2003 and whether this mechanism is different for varying causes of death. We obtained daily counts of total and cause-specific mortality for Essen, Germany, for the years 2000-2006. We used time-series regression methods to separate a possible additional effect of sustained heat from the temperature effect and included air pollution, influenza epidemics, long-term and seasonal trends, days of week and bank holidays as covariates. The maximum daily relative risk of all-cause mortality during the heat wave was 1.28 (95% CI 1.06-1.53). The maximum relative risks of cardiovascular and neoplastic mortality were 1.25 (95% CI 0.95-1.65) and 1.35 (95% CI 1.00-1.82), respectively. The effect on respiratory mortality was delayed; the maximum relative risk was 1.66 (95% CI 1.19-2.23) 6 days after the heat wave. We found that periods with sustained heat especially affected respiratory mortality, whereas for cardiovascular and neoplastic mortality no distinct influence could be shown.