The diet of a neotropical fish species (Creagrutus bolivari) was studied using different methods: the Cortés 1997 graphic method and two relative importance indexes,one designed by George and Hadley in 1979 (% RIi) and another proposed by Cortés in 1997 (% IRIi). Only discrete prey (aquatic and terrestrial invertebrates) were used in the diet analysis. The occurrence (% A), number (% N) and weight (% G) frequencies were estimated for each food item. A total of 240 specimens (size range from 25.3 to 64.7 mm standard length) were analyzed. For values of % A > 50 %, % RIi was higher than % RIi with an opposite trend (up to 130 times) for values of % A < 50 %. % IRIi overestimated the relative importance of prey for high values of % A, while it underestimated the relative importance of prey for low values of % A, showing a nonlinear relationship to % A. The variables used in the Cortés graphic method are insufficient to evaluate predator feeding strategy (generalization versus specialization). Therefore, we suggest including other variables, such as population and individual trophic niche breadth and a suitable field sampling record. An extended sampling record is necessary to classify prey as rare or dominant according to Cortés’ classification. The indexes of relative importance and graphic method are complementary. Compound indexes that comprise mass, number, and occurrence provide a more accurate description of diet and allow comparative studies. The graphic analysis is easy to interpret and helps in the evaluation of prey importance and predator feeding strategy.