Broad survey census data for mantled howler monkeys, Alouatta palliata, from La Selva biological Reserve, northeastern Costa Rica, are analyzed with an agglomerative hierarchical clustering method to estimate minimum number of troops. The estimate of 15 troops (7-15 howlers/km2) provided by the duster analysis is similar to an independent census of the same population using the triangulation technique. These data indicate the potential usefulness of duster analysis methods to estimate minimum number of primate troops within an area. Cluster analysis has several advantages over more traditional primate census techniques: the assumption that it is necessary to achieve reliable estimates from line transect sampling is not required (i.e., points directly on the line will never be missed); census data can be collected by one person making it easier logistically than conducting a census with the triangulation technique which requires many people; census data collected over a brief period of dme can be used to provide quantitative estimates of population density; and observations provided by others in the survey area during the time period of the census may be used. Cluster analysis is recommended when attempting to estimate population densities of primates in tall evergreen forests where visibility is poor, when animals have long inactive periods, few or unpredictable vocalizations, and/or occupy the top of the canopy. The duster method of analyzing broad survey data may also be applicable to other territorial mammalian populations that are difficult to census by other means.