The effect of token reinforcement on the verbal participation of long-stay male psychiatric patients in a social activity was studied in four experiments. In Experiment I the characteristic effects of continuous reinforcement were found with 21 of the 37 patients on the ward. In Experiment II an effective and quick way of establishing tokens as conditioned reinforcers was demonstrated with 25 patients. In Experiment III the 37 patients were divided into four groups, with high responders, low responders and non-responders equally distributed in each, to determine the effect of fixed ratio schedules of token reinforcement on verbal participation. In Experiment IV the effect of different schedules of variable ratio token reinforcement on the same group of patients is examined. To introduce the studies a brief but comprehensive survey of token systems in applied settings is made. The term 'token economy' is defined and distinguished from 'token reinforcement system'. Studies of non-verbal behaviours using intermittent token reinforcement are reviewed. This is followed by a review of studies of verbal behaviour in which token reinforcement has been variously delivered on continuous, fixed interval, variable interval, fixed ratio and variable ratio schedules. Finally a general survey of the operant conditioning of verbal behaviour is made and a comprehensive bibliography of studies with token systems until December 1975, is provided.