A validity test of a Food Frequency Questionnaire was carried out using 50 students of health occupation in São Paulo, Brazil. Therefore, a three day dietary record was used as reference method and variables such as energy, macronutrients and dietary fiber were analyzed. The accordance between the Food Frequency Questionnaire and average data from dietary record was tested with kappa statistics and intra-class correlation coefficients (ICC). Limits of agreement were estimated by the Bland-Altman's method. Better results were found for calories (ICC 0.43; 95%CI 0.17 - 0.63) and non-energy-adjusted nutrients, except dietary fiber (ICC 0.34; 95%CI 0.07 - 0.56). The percentage of individuals classified in the same category of consumption was nearly half (49.8%), while only 16% of them were classified in opposite categories. With the exception of lipids, other analyzed variables tended to be overestimated by the Food Frequency Questionnaire. The Food Frequency Questionnaire is recommended as a method of assessing food intake of university students in studies which focuson calorie estimates and also intend to classify groups into intake categories.