Conservation and ecological monitoring programmes often estimate animal densities over time, but in wooded and forested areas practical techniques are still poorly developed. Here I have examined five simple methods of deriving densities of large and medium-sized mammals using line transects driven through miombo woodland habitat in Africa. These methods calculated area by dividing the number of individuals seen by (i) an average of each species' sighting distances, (ii) a fixed 200 m belt width, (iii) the area visible from the centre of the transect, (iv) visible area weighted by species' vegetation preferences, and (v) by dividing the number of groups seen by area visible from the transect. Individual-based methods produced quite divergent estimates of species' densities and overall biomass with belt transects giving the lowest values. The group method and corresponding individual-based method gave similar values, however. Most of these calculations yielded considerably higher density estimates than aerial surveys conducted in the same area over a similar time period. Sightings of species were not distributed evenly across vegetation types although the majority of mammal species were observed more often in open habitat. These findings indicate that ground-based conservation monitoring programmes should set transects through several vegetation types, restrict comparisons to those studies that use the same methodology, and refrain from comparing ground and aerial surveys.