Routine biomonitoring studies that compare abundance of benthic macroinvertebrates upstream and downstream from contaminant discharges generally cannot be used to demonstrate causal relationships between stressors and biological responses. In this study I describe stream microcosm and field experiments designed to support results of a 12-year monitoring project in the metal-polluted Arkansas River, Colorado. Microcosm experiments established concentration-response relationships between heavy metals and several structural (abundance, richness) and functional (macroinvertebrate drift, community respiration) endpoints. EC 10 values, the metal concentration that reduced abundance or richness by 10%, were generally lowest for mayflies and stoneflies, indicating greater metal sensitivity for these groups. Macroinvertebrate drift and community respiration showed highly significant concentration-response relationships with heavy metals and were generally more sensitive than structural measures. Potential interactions among metals were investigated by comparing community-level responses to Zn alone, ZnϩCd, and ZnϩCuϩCd. Results showed that macroinvertebrate responses to a mixture of three metals were generally greater than responses to either Zn alone or ZnϩCd. Long-term monitoring in the Arkansas River (Colorado) before and after remediation of metal inputs provided correlative evidence that elevated heavy metal concentrations altered community composition and reduced abundance of metal-sensitive organisms. However, these field data were not especially useful for estimating safe concentrations of heavy metals that would be protective of benthic communities. Microcosm and field experiments provided further support for the hypothesis that metals caused alterations in benthic community structure and provided more precise estimates of safe metal concentrations.