Epidemiological investigations on nutrition and chronic diseases require high reliability, accuracy, low cost and feasibility of the system through which dietary information is collected. Food frequency questionnaire are commonly used in large epidemiological investigations. However, repeated 24 hour recall should be considered to integrate this information. Multiple 24 h recalls in random samples of large cohorts appear to be convenient choice to improve data accuracy. This procedure also allow calibration of dietary information in case of pooling of data from different cohorts. Multiple 24 h recalls can be easily carried out through telephone interviews instead of multiple face-to-face. This reliable procedure, tested in a study reported in this paper, may improve compliance.