Introduction: The gold standard to estimate individual usual intake is the mean of multiple day's intake. This estimation procedure is based on Beaton's model, who consider independence and two sources of variation: between and within subjects. There are evidences and physiological arguments against validity of Beaton's hypothesis, this fact could have relevant effects in dietetic assessment. Subjects and Methods: A food consumption of 43 middle class university women was recorded during 28 days. They were trained in order to assure quality records. Results: Low values in energy intake were reported, and no differences between energy intake during weekend and weekdays was observed. Considering reported intakes for each person as a time series, no autocorrelations was observed considering lags within 1 to 7 days. Differences in within subjects variances was observed. Conclusion: Deficiencies in subject's energy intake corresponds to disorders in university students' nutrition, such deficiencies could be the explanation of the observed independence in daily energy intake despite evidences of the existence of correlations reported in different papers. The hypothesis of variance homogeneity was not true, so it was not correct to talk about a "within variance", instead a population's probability distributions for variances was considered, as a consequence Beaton's model must be modified. A probability distribution to characterize the number of days needed to estimate energy intake for subjects in population was considered.