Studying elusive species of conservation concern might be difficult for technical and ethical reasons. However, censuses can be based on the observation of activity indices. When coupled to non-invasive genetic methods this approach can provide extremely precise information about population size, individual movements and diseases. However, the design of optimal sampling is dependent on a knowledge on group distribution and possible variations of detectability of index targets. The aim of this study was to document the distribution of Yunnan snub-nosed monkey indices in space and time in that perspective. Based on transects carried out across the range of a fed population and on counts along the trail across the range of a wild group, we show that 2-3 day stays of a group in a place of some hectares were sufficient to get an homogeneous distribution of indices. Furthermore, the number of indices found were dependent on both pig presence and season. On the other hand, on a large scale of 100 km(2) indices were spatially distributed as nested clusters. Indices distribution indicated a strong preference towards southern slopes and altitudes ranging between 2900 and 3400 m. Those observations pinpoint the importance of considering spatial scale to organise sampling designed to estimate population distribution.