Research into social skills training has, in common with much other contemporary educational psychology practice, been beset by the problem of how best to evaluate effectiveness. Alongside the movement towards more behaviourally inclined approaches has grown a dissatisfaction with the traditional methods of evaluting effectiveness. If we recognize behaviour per se as being the appropriate level for intervention, whether in relation to academic skills teaching, classroom management, social skills training or whatever, then it follows that effectiveness can only logically be assessed in terms of measurable changes in observed behaviour. Evaluation methodology has tended to lag behind interventional methodology, however, so that it has not been uncommon to see behavioural work assessed in terms of changes in attitude and/or knowledge. On occasion evaluation is not even attempted and behaviour change is assumed to generalize outside the specific confines of the intervention situation.