The aim of this study was to test for a correlation between party size and food (fruit) availability among the M group chimpanzees (Pan troglodytes) in the Mahale Mountains, Tanzania. Chimpanzee unit groups (or communities) show fission-fusion grouping patterns and form temporal parties. Fruit availability is assumed to be one of the important limiting factors in relation to the size of these parties. Different methods have been proposed to measure party size, but they all appear to focus mainly on two aspects of grouping phenomena. In "face-to-face parties", party size is measured by scan sampling, whereas in "nomadic parties", all members observed during a specific time period are counted. The mean monthly group size resulting from these two measures was compared with fruit availability, i.e. fruiting plant density and mean potential patch size. Nomadic party size was correlated with both values. Thus, party formation at this level was considered to be sensitive to overall fruit availability in the habitat. On the other hand, face-to-face party size remained stable and showed weak or no correlations with density and potential patch size. Although large patches are available during the peak fruiting season, Mahale chimpanzees depend on the liana species Saba comorensis, which, when fruiting, encourages individuals to spread out to eat. Thus, the lack of correlation between face-to-face-party size and fruit availability was attributed to the influence of physical limitations countervailing the fluctuation in fruit availability. Maximum face-to-face party size relative to unit-group size, regarded as the cohesiveness of a unit group, was compared among sites. The values differed largely: Mahale groups M and K, Bossou, and, in some years, Budongo, showed high cohesiveness, while others remained low. Thus, the distribution of the most important food during the fruiting season in each study site may be a crucial factor in the grouping phenomena of chimpanzees.