The comparison of precision is often advocated for the selection of an appropriate census and/or monitoring method for wildlife, but little attention is generally paid to their cost effectiveness, a crucial criterion given budgetary and logistical constraints. We present six direct count methods conducted in a communal area of the Zambezi Valley, Zimbabwe, and compare them in terms of (1) effort and cost to survey an area (sampling efficiency), and (2) efficiency in data collection (detection efficiency). Methods ranged from c.US$0.2 to over US$6.0/km2 and needed from 0.1 to 5.0 human-h/km2. The comparison of efficiencies showed the advantages of simple ground methods: foot counts and particularly bicycle counts appear well adapted to the ecological and human context of our study. The relative benefits and constraints of the different methods are discussed in the context of a community-based wildlife management programme. Keywords: African mammals, census techniques, cost efficiency, non-protected area.