Validation studies of food frequency questionnaires (FFQs) describe the extent to which the FFQ reflects true diet and the relation between measured and true diet (calibration). Calibration data can be used to estimate the relation between disease and diet that would have been observed in the absence of error due to the FFQ. In this paper, the authors consider the optimal design of a validation study when the goal is precise calibration of an FFQ. The authors posed the following question: Under the constraint of a fixed total cost for a validation study, what is the optimal choice of number of subjects (n) and number of days (m) of diet records (or 24-hour recalls) per subject? The optimal n and m were found to depend upon 1) the ratio between the costs of the initial and subsequent 1-day diet records and 2) the ratio of the variance in day-to-day nutrient intake to the variance in true diet for a fixed FFQ value. Data for the two ratios and optimal values of n and m are given under a variety of realistic scenarios. The authors conclude that in most settings the optimal study design will rarely require more than four or five 1-day diet records per subject.