The paper reports on a series of field experiments based on a non-standard common-pool resource model and carried out with communal farmers in southern Africa. We frame an experimental design used by Cardenas et al. (2008) in Thailand and Colombia according to the grazing situation in semi-arid regions. We analyse the efficiency of two simple, imperfectly enforced property-rights mechanisms that base on both, informal realworld arrangements practised in the communal areas as well as on the procedure pursued by the Namibian government, where mostly those farmers with big herd sizes get granted access to resettlement farms. Our results suggest that in the short run, the introduction of property rights increases the economic returns and has positive ecological effects, but that these effects diminish in the long-run if imperfectly enforced. It further seems that players’ propensity to co-operate is constant across the experiment: farmers who behave less co-operatively in an open access situation do so as well when they get granted exclusive property rights.