Unsustainably high mortality within the first 2 years of life prevents endangered Lost River Suckers Deltistes luxatus in Upper Klamath Lake, Oregon, from recruiting to spawning populations. Massive blooms of the cyanobacterium Aphanizomenon flos‐aquae and their subsequent death and decay in the lake (bloom‐crashes) are associated with high pH, low percent oxygen saturation, high total ammonia concentrations, and spikes in the cyanotoxin microcystin. Poor water quality within the lake is considered the most likely cause of juvenile sucker mortality, but mechanisms causing the high mortality are not known. We introduced PIT‐tagged age‐1 Lost River suckers into three continuously monitored mesocosms in Upper Klamath Lake to determine the timing of juvenile sucker mortality relative to pH, temperature, and dissolved oxygen. Mortality was inferred from a lack of movement detected on remote PIT tag detection equipment within each mesocosm. Mortality was compared among mesocosms and an indoor tank‐held control group. We fitted time‐varying Cox hazard models to test hypotheses about short‐term and chronic effects of single and co‐occurring water quality parameters on the daily hazard rate. Presumed healthy or moribund fish that were collected pre‐season, mid‐season, or at the end of the study were examined macroscopically and histologically to generate inferences about the causes of mortality. Models did not indicate a plausible association between water quality variables and mortality. Hypoxia preceded periods of higher mortality at two of three sites but did not co‐occur with mortality. Hatchery‐reared Lost River Suckers confined to mesocosms may not represent the behavior of wild fish, and it is unclear whether the same factors affect the mortality of wild age‐0 suckers.