During the past decade, the behavioral needs, and especially the social needs, of laboratory animals have received increasing attention. New captive care guidelines have been developed, and these advocate group housing of laboratory animals whenever possible and appropriate. Analyses of behavioral observations are commonly used to assess the effects of experimental manipulations on behavioral responses. In studies of animal welfare, stress levels and effects on well being can be measured in this manner. Collecting the proper amount of data both to allow statistical analyses and to optimize time investment in data collection is a practical concern in behavioral research. The aim of the present study was to develop a simple method to estimate how much behavioral data should be collected in order to achieve a preset level of reliability across observations. This paper examines the behavior of 12 rhesus monkeys (Macaca mulatta) housed in two small social groups. Each monkey was observed for a total of two hours in 10 minute observation periods spread over 12 days. Accumulated Means Analysis, in which the accumulated mean for a behavior across successive observation sessions must meet several criteria, is proposed as a method to assess reliability across observations, thereby providing information concerning the optimum number of observational data sessions that need to be collected. The data in this study indicate that this optimum number of observations varies with the behavior being measured and with the group observed, as does the value of the Accumulated Means Analysis technique. Six hours of observation of rhesus macaques may be sufficient to provide the type of simple, yet reliable time budget (for a specific window of time) needed by those managing groups of primates. Accumulated Means Analysis should be applicable to behavioral data collected during multiple time periods on other non-human primate species, both in captivity and in the field.