The dataset contains the results of surveys on small mammal communities conducted in various habitat types across Romania, starting with 2000. Location of sampling sites, habitat characteristics, trapping effort and the number of individuals captured as well as the capture index for each species are given.

We live-trapped small mammals using artisanal wooden box-traps (18 x 10 x 8 cm) or Fitch traps (40 x 6 x 6) set in transects (trap-lines) or grids, mostly during the warm season. Transects included 30 to 40 traps set 15 m apart along the contour lines and parallel to the closest watercourse, forest edge, road or trail, within homogenous forest habitats; ecotones were avoided. Trapping grids included between 6 x 4 and 8 x 8 traps at distances of 10 or 15 m. 

Traps were baited with sunflower seeds and apple slices, but no prebaiting was done. Traps were checked in the morning and at dusk for two or three consecutive days. Because many traps were disturbed by weather, animals or people, the trapping effort differed greatly among transects.

We identified captured individuals to species based on morphological traits, marked them by fur clipping, and then released each at its trapping site. Recaptures were not included in the dataset.

For each transect or grid we estimated a series of habitat variables generally considered to be important for small mammal populations: percent cover of tree canopy (referred to as tree cover hereafter) also recorded separately for coniferous trees, percent of shrub and herbaceous layers, mean height (cm) of herbaceous layer, distance (m) to closest watercourse, river or creek, and elevation (m). Tree cover was estimated by the percent of the ground where light fell directly; distance to water and elevation were measured for the trap placed at the center point of the transect. Soil moisture, abundance of rocks, coarse woody debris and human disturbance were estimated as ordinal variables. 

Additional details on the variables are given separately for each subset of the dataset in the corresponding read.me files.