Altitudinal gradient often leads climatic effects which would result from latitudinal changes, and so has been used as a surrogate for latitudinal gradient. This has prompted the research on biodiversity dynamics across an elevation gradient. Little is known, however, about the response of soil biota to the changes of climate, vegetation and soil along an elevation gradient. We examined the changes of soil faunal diversity under the four typical vegetations i.e. evergreen broadleaf forest (EVB), coniferous forest (COF), dwarf forest (DWF), and alpine meadow (ALM) during April 2007 to 2008 in the Wuyi Mountains in southeastern China. Our findings were as following: (1）The composition of dominant groups were found no difference along the elevation gradient with Mites and Collembolans both accounting for 60% of the total. The common groups were Eiymenoptera, Diptera, Coleoptera, Symphyla, Isopoda, and Araneae, respectively, at the four sites. The Jaccard comparability coefficients (between 0.8000 and 0.8696) of soil fauna were similar at three forest sites while there were distinct different between sites of EVB and ALM, and COF and ALM, indicating that the aboveground vegetation had significant effects on the soil faunal composition. (2) The total number, density and diversity of soil fauna decreased with the increasing above sea level along the elevation gradients. The total fauna number (17) and density at ALM (25925.4 individual /m2) were less than those at EVB (24 and 56195.9 individual /m2). Density-Group (DG) Index at the four sites ranked as EVB（6.238）＞COF（4.015）＞DWF（2.124）＞ALM（1.754）(P<0.05). The abundance of dominated groups e.g. Mites (F=9.25, P < 0.01), Collembolans (F=5.93, P < 0.05), Eiymenoptera（F=6.35, P < 0.001), Coleoptera (F=4.42, P < 0.05), and Symphyla (F=2.96, P < 0.05) descended across the elevation gradients. (3) Soil fauna mostly distributed at the surface soil layer across the elevation gradients. The individuals of Eiymenoptera, Diptera, Collembolans, and Isopoda, however, mostly assembled in 10-25cm soil at ALM, which might be derived from some soil faunal groups moving from surface to deeper layer, avoiding the unfavorable climate surroundings at higher altitudes. (4) Temporal patterns of soil faunal number and individuals fell in two groups across the elevation gradient: (a) Peak values of the soil fauna were in the autumn at sites of EVB and COF, (b) Individuals of the soil fauna mostly distributed in the summer at sites of DWF and ALM. Our results indicated that the structure and makeup of the faunal community might be important response processes along an elevation gradient in the Wuyi Mountains. The number, density, and diversity decreased significantly, which supply some data for soil faunal distribution study in subtropics and some aides in understanding of soil faunal response to altitude changes.