A laboratory study was performed which compared structural responses in gill tissues with fry survival and growth in rainbow trout salmo gairdneri exposed to sub-lethal concentrations of oil shale retort water for 60 days. Histological techniques included tissue fixation in buffered glutaraldehyde followed by buffered osmium tetroxide and embedment in Araldite 502. Sections 1-pm thick were cut from tissue blocks so that each section contained a sample of several filaments, each cut form a different level along the filament. With high-resolution light microscopy, the containment response was quantified using white blood cell infiltration, chloride cell and mucous cell proliferation, mitotic index, and respiratory diffusion distance. The no-observed-effects-concentration (NOEC) was 0.009% retort water for survival. In contrast, the NOEC for quantified histological parameters was 0.001% retort water for fry growth and 0.004% retort water for survival. In contrast, the NOEC for quantified histological parameters was 0.001% retort water for white blood cell infiltration and 0.002% for four other structural parameters. Thus, gill histopathology was more sensitive than traditional chronic toxicity endpoints. We conclude that quantitative histopathology may be used to predict "safe" levels of at least some environmental contaminants.