I investigated the utility of environmental variables in helping to predict (1) the presence of self-sustaining trout populations and (2) catch per unit effort (CPUE) of trout in 53 lakes of Yosemite National Park, California, from June to October, 1996. Rainbow trout (Oncorhynchus mykiss irideus), brook trout (Salvelinus fontinalis) and brown trout (Salmo trutta) were caught by gill net in 47 lakes and aged using otoliths to determine the presence of self-sustaining trout populations. Stepwise logistic regression analysis identified total visible spawning habitat area, the presence of brook trout, maximum depth, water temperature, and distance from trailhead as significant variables for predicting the presence of self-sustaining rainbow trout populations. Total visible spawning habitat area and maximum depth were significant for predicting the presence of self-sustaining populations of any trout species. Stepwise multiple regression analysis identified the presence of brook trout, total visible spawning habitat area, and water temperature as significant variables for predicting rainbow trout (CPUE), while total visible spawning habitat area was significant for predicting trout CPUE (pooled for all species caught). The most accurate model constructed from regression analyses predicted 96.2 percent of the distribution of self-sustaining rainbow trout populations observed by the author. The utility of CPUE as an indicator of self-sustaining trout populations is limited, however, since the best model constructed using environmental variables only explained 32 percent of the variation in rainbow trout CPUE.