The results of comparisons of behavioral data between individuals, age–sex classes, seasons, sites and possibly even species may depend on sample size and the computational method used. To establish whether these are valid concerns, we compared results for percentage time spent feeding on major food types (fruit, leaves, flowers, invertebrates, bark, pith and other) for two orangutan populations in Sabangau (24 months) and Tuanan (29 months), Indonesian Borneo. Both the minimum follow limit included in analyses and the computational method used produced small, but statistically significant, differences in the results obtained, and the differences were more common for food types eaten less frequently. In addition, using different computational methods produced more significant differences than did including different minimum follow lengths in analyses. The computational method used also influenced the results of tests for differences in diet composition between age–sex classes. Thus, the method used can influence the results obtained and, hence, it is important to state explicitly the minimum follow limit included and computational method used to compile averages, and to ensure standardization in methods when comparing data between age–sex classes, time periods, field sites or species. Am. J. Primatol. 71:353–358, 2009. © 2008 Wiley-Liss, Inc.