The diet diary technique can produce reliable and valid records of the food and fluid intake of free-living humans. Procedures to maximize accuracy by motivating subjects, verifying records and minimizing bias are described. Underestimation and random error may tend to obscure relationships but does not artifactually produce them. Meal pattern analysis of the diary records relates the meal size, composition and intermeal interval to preceding conditions with univariate and multivariate regression techniques. Although causation cannot be conclusively demonstrated, the analyses can test for potential causal influences. It is concluded that the diet diary technique is the method of choice for investigations of the ingestive behaviours of free-living humans.