Large ungulates critically influence forest structure and functioning besides being seriously threatened by anthropogenic pressures. For assessing their populations, surveys of ungulate sign encounters are widely used because of their practicality. However, these yield unreliable results because of their failure to address the problem of imperfect detection. Here, we present an innovative application to address this key weakness in traditional ungulate sign surveys. We describe the ecological process of ungulate sign deposition as well as the observation process of sign detection in our modelling. We simulate 183 ecological and sampling-related parameter values to first evaluate model performance. Simulation results demonstrate that we can achieve good estimates of animal density when the radius of the animal daily movement range is accounted for during survey design. We design and conduct a field survey of ungulate signs to estimate ungulate densities using both occupancy and distance sampling approaches. For five species of ungulates, the densities estimated from our sign survey (number of ungulate clusters km−2) were 1.46(0.68) chital Axis axis, 1.42(0.67) sambar Rusa unicolor, 1.01(0.44) gaur Bos gaurus, 0.74(0.39) wild pig Sus scrofa and 1.42(1.59) muntjac Muntiacus muntjak, and were similar to those generated from line transect sampling 2.16(0.76) chital, 2.47(0.56) sambar, 0.94(0.3) gaur, 1.09(0.37) wild pig and 4.03(0.83) muntjac), except for muntjac. The potential utility of this approach extends beyond sign surveys of forest ungulates to a wider range of animal monitoring contexts, including those based on scent-station surveys and camera trap surveys of elusive mammals.