There are several approaches aiming to evaluate nutritional status available for epidemiological studies. Dietary methods are some of the most recommended because of their ability to detect nutritional deficiency in its initial stage. Currently, the Semiquantitative Food Frequency Questionnaire (SFFQ) has been chosen as the best dietary method, especially when dealing with chronic degenerative diseases. This tool, like many others, is subject to intrinsic variability and measurement errors. However, both phenomena can be identified and somehow treated through different methodologies such as comparison of dietary means, reliability studies, relative validity and calibration. In a study carried out in the town of Bambuí, nutrient intake measurements were obtained using SFFQ with photos and the 24-Hour Recall (R24) questionnaire with and without foodstuff replicas. Average nutrient intake obtained by the SFFQ was significantly higher compared to R24 questionnaires for all nutrients. However, when the averages were adjusted for total caloric intake, means were similar, except for fiber, zinc, vitamins C, B6 and E, and cholesterol consumption. The calibrated averages obtained by SFFQ were generally lower than those from the non-calibrated SFFQ and significant for lipids; fiber; zinc; vitamins C, E and A; saturated, monounsaturated and polyunsaturated fatty acids intake.