To understand the species diversity distribution pattern of Gongga Mountain, we surveyed the diversity of non-volant small mammals along the elevational gradient on the eastern slope of Gongga Mountain from April to September in 2010. Eight sampling sites were set along the elevational gradient from 1200m to 4000m with an interval of 400m, and 701 small mammals were captured in 28800 trap nights. Plusing the 25 field observed squirrels, our study surveyed totally 726 individuals representing 25 species that were belong to 16 genera, 6 families and 3 orders. The two non-parametric estimations of species richness Chao2 and Jackknife2 as well as the species cumulative curve demonstrated that the sampling of non-volant small mammal species were adequate and well reflected the diversity distribution pattern along the elevational gradient. The results showed that the species richness and abundance patterns for non-volant small mammals along the eastern slope of Gongga Mountain were hump-shaped with the highest richness at mid-elevation, but decreased at both lower and higher elevations. In contrast, species evenness displayed a distribution pattern with high evenness at high and low elevations but low evenness at mid-elevation, whereas species dominance index increased gradually along the elevational gradient. Four widely-used diversity indices (Shannon-Wiener, Simpson, Fisher-α and Margalef) were used to compare the diversity discrepancy between different elevational communities and also used to reveal the diversity distribution pattern. The Fisher-α and Margalef diversity indices seemed to reveal the diversity discrepancy between different communities much better than the Shannon-Wiener and Simpson diversity indices. Meanwhile, Simpson diversity index failed to reflect the unimodal distribution pattern of diversity along the gradient. The elevational distribution patterns of the diversity indices showed the same distribution pattern as the species richness and abundance with peaks at the mid-elevation. This result was also supported by clustering analysis based on the species composition of the 8 elevational bands. This unimodal diversity distribution pattern suggests that the mid-elevational area should be paid more attention in the diversity conservation and ecological management of Gongga Mountain, because it harbors the highest diversity of small mammals and these irreplaceable species play important and fundamental roles in maintaining the health of mountain ecosystems. Furthermore, standard and unified sampling methods could be contributive to the integration of research data so as to reduce any artificial errors from different diversity investigation methods.