The Reference Condition Approach to bioassessment has been used in many contexts of environmental assessment and monitoring, employing biota descriptors from predicted occurrence of individual taxa (e.g. AUSRIVAS, RIVPACS), to compositional indices (e.g. BEAST), to structural descriptors (e.g. GIMRCA). Without known levels of ecosystem degradation, however, it is impossible to evaluate and compare the power of these different approaches to detect ecosystem response to stressors. I used real environmental and biota data from a bioassessment study of streams in northern British Columbia, together with a stochastic simulation programmed in SAS, to assess the effect of i) degree of initial deviation from Reference condition (none, small, large), ii) type of stressor ( metal effluent, nutrient augmentation, increased suspended sediments), iii) intensity of exposure to stressors (none, low, moderate, high), and interactions of these factors on the power of RCA to detect ecosystem degradation using both compositional and structural descriptors.