Three separate food grouping systems were derived for diet evaluation by utilizing a computer clustering program. System 1 was based on grouping of foods according to nutrient quality and quantity, system 2 was based on nutrient quality and frequency of consumption and system 3 was based on nutrient quality and kilocalories consumed. Defendable nutritional principles as well as knowledge of present nutritional concerns and eating habits of the U.S. population underpinned the derivation of the three systems. Over 4,000 individual foods were condensed to approximately 325 groups. The three food grouping systems which were developed were compared for accuracy by calculating nutrients for a mean daily food record for a random sample of 126 adults in the National Food Consumption Survey. Percent error for each of 15 nutrients was calculated for each derived system as contrasted to the value which resulted from using individual nutrient values for discrete foods. In evaluating the 3 systems the most accurate comparison resulted from the system which was based on frequency of consumption. The results of the comparison indicated that 11 of the 15 nutrients which were determined were consistent in all systems. The exceptions were vitamin A, vitamin B12, riboflavin and ascorbic acid. These inconsistencies can be attributed to the nature of the food sources and the nutrients themselves. It was demonstrated that fortified foods and food supplements which contain varying amounts of selected nutrients require individual treatment for accurate analysis. These results supported the hypothesis that a shortened dietary analysis system can be used without compromising accuracy.