In recent years the importance of facilitative interactions in ecological communities is increasingly recognized. This phenomenon has been observed repeatedly, particularly in vegetation communities, in a wide range of environmental conditions. The current hypothesis predicts that the role of facilitation becomes increasingly important in conjunction with increasing stress. Several empirical studies, however, failed to detect such patterns, particularly at the extreme ends of the stress gradients. Herein, we present a conceptual model that may resolve discrepancies between expected and observed and provides a more precise framework of the existing hypotheses. By relaxing two common assumptions commonly used by the stress-gradient hypothesis (SGH) we are able to demonstrate that under some circumstances the importance of facilitation may be less at the extreme ends of these gradients. Namely, we first re-emphasize the notion that physiological response is not linear with respect to environmental changes along stress gradients. Second, it is argued that the net outcome of facilitative and competitive interactions is reflected in the fitness of individuals as a product of these two processes, in contrast to the commonly applied assumption of additivity. Accordingly, a synthesis of the concepts of population biology (measures of fitness) and plant physiology (nonlinear responses) with the stress gradient hypothesis while retaining the original simplicity of the SGH model contributes to a better specification of the predictions of the stress-gradient hypothesis and the resolution of observed contradictions.