This paper proposes a hypothesis that relates biodi- versity, community biomass, and ecosystem func- tion to a gradient of stress. According to this hy- pothesis, biodiversity has a low threshold of response to stress, whereas biomass and function are stable or increase under low to moderate stress and decrease only under high stress. This hypothe- sis was tested by examining communities of pri- mary producers in streams under stress from mine drainage in the Rocky Mountains of Colorado, USA. Mine drainage exerts chemical stress (low pH, dis- solved metals) as well as physical stress (deposition of metal oxides) on stream biota. Diversity of pri- mary producers was usually more sensitive to stress from mine drainage than community biomass (chlorophyll a) or primary production. Diversity was negatively related to all stresses from mine drainage, but it was especially low in streams with low pH or high concentration of dissolved zinc. Biomass and production were high in streams with only chemical stress, but they were often low in streams with physical stress caused by metal oxide deposition. Stream sites with aluminum oxide dep- osition usually had very little algal biomass. The rate of metal oxide deposition, presence of alumi- num oxides, and pH together explained 65% of the variation in biomass. The rate of net primary pro- duction was highly correlated with biomass and had a similar response to stress from mine drainage. Overall, chemical stresses (low pH, high concentra- tion of zinc) generally led to the hypothesized trends in our model of ecosystems under stress. Physical stress (deposition of metal oxides), how- ever, led to variable responses, and often decreased biomass and function even at low intensity, con- trary to the original hypothesis. Thus, the nature of ecosystem response to stress may differ for chemical and physical stresses