Inbreeding depression (ID) is generally assumed to increase under stressful conditions, but a number of studies have found the opposite pattern, i.e. that crossed offspring were more capable of exploiting benign conditions. Alternatively, the phenotypic variation hypothesis predicts that not stress intensity, but enhanced phenotypic variation in an environment leads to increased ID. We subjected inbred and crossed offspring of Silene vulgaris to drought, simulated herbivory, copper contamination, and two levels of nutrient deficiency and shade. In contrast to the predominant expectation, most stress treatments decreased inbreeding depression. With increasing nutrient limitation, ID decreased strongly, whereas under increasing shade ID did not change. These differences may be due to purging in the population of origin where conditions are nutrient-poor and dry, but not shaded. In contrast to the greenhouse experiment, ID was higher in a field site than in a more benign common garden. However, the predictions of the phenotypic variation hypothesis were met in both the greenhouse and the field vs. garden experiment. The results suggest that there may be no general relationship between ID and stress intensity, but specific effects of stress type and the novelty and variability of the environment.