Forty-eight children from grades 3 to 6 of two inner city primary schools were selected on the basis of poor performance on a social perception test. Subjects were randomly assigned to one of three experimental conditions: social perception training (SPT); attention placebo control (APC; drama activities) or a no treatment control (NTC; assessment only). The SPT and APC groups participated in nine twice weekly sessions over a five week period. Children receiving SPT did not show a significantly greater improvement in social perception skills than either of the control conditions. There was also no significant effect of SPT on certain measures of behavioral functioning which have previously been associated with social perception skill, namely peer sociometric status, Walker Problem Behaviour Checklist or Childrens' Depression Inventory. The findings are discussed in terms of implications for future research.