Abstract The phenotypes of plants can be influenced by the environmental conditions experienced by their parents. In some cases, such parental effects have been found to be adaptive, which has led to much speculation about their ecological and evolutionary significance. However, there is still much uncertainty about how common and how predictable parental environmental effects really are. We carried out a comprehensive test for parental effects of different environmental stresses in the model plant Arabidopsis thaliana . We subjected plants of three Arabidopsis genotypes to a broad range of biotic or abiotic stresses, or combinations thereof, and compared their offspring phenotypes in a common environment. The majority of environmental stresses (16 out of 24 stress treatments) caused significant parental effects, in particular on plant biomass and reproduction, with positive or negative effects ranging from −35% to +38% changes in offspring fitness. The expression of parental effects was strongly genotype-dependent, with some effects only present in some genotypes but absent, or even in the opposite direction, in others. Parental effects of multiple environmental stresses were often non-additive, and their effects can thus not be predicted from what we know about the effects of individual stresses. Intriguingly, the direction and magnitude of parental effects were unrelated to the direct effects on the parents: some stresses did not affect the parents but caused substantial effects on offspring, while for others the situation was reversed. In summary, parental environmental effects are common and often strong in A. thaliana , but they are genotype-dependent and difficult to predict. Significance Stress experienced by plants can alter the phenotypes of their offspring. To understand the ecological and evolutionary significance of such parental effects, we must know how common and how predictable they are. In a large experiment with Arabidopsis thaliana , we show that the majority of 24 environmental stresses cause significant, and often strong, positive or negative parental effects. However, we also find that parental effects are genotype-specific and unrelated to the direct effect of individual stresses, and that multiple stresses often act in non-additive ways across generations. Thus, parental effects appear to be common and strong, but difficult to predict. Our findings have important implications for the study of plant responses to environmental change, and the design of stress experiments.