Harsh conditions (e.g., mortality and stress) reduce population growth rates directly; secondarily, they may reduce the intensity of interactions between organisms. Near-exclusive focus on the secondary effect of these forms of harshness has led ecologists to believe that they reduce the importance of ecological interactions, such as competition, and favor coexistence of even ecologically very similar species. By examining both the costs and the benefits, we show that harshness alone does not lessen the importance of species interactions or limit their role in community structure. Species coexistence requires niche differences, and harshness does not in itself make coexistence more likely. Fluctuations in environmental conditions (e.g., disturbance, seasonal change, and weather variation) also have been regarded as decreasing species interactions and favoring coexistence, but we argue that coexistence can only be favored when fluctuations create spatial or temporal niche opportunities. We argue that important diversity-promoting roles for harsh and fluctuating conditions depend on deviations from the assumptions of additive effects and linear dependencies most commonly found in ecological models. Such considerations imply strong roles for species interactions in the diversity of a community.