The practice of training evaluation has not kept pace with prescription, and evaluations being being done are frequently negligent of appropriate controls needed to draw valid conclusions. A comparison was made of training outcomes contrasting results obtained using carefully controlled scientific approaches with those from a more popular less scientific approach. The research design involved the collection and analysis of data from a single organizations managerial training program. , Three different methods of training evaluation were studied: an "immediate reaction" rating sheet, a self-report participant survey, and a similar survey completed by t he participants' subordinates. Bo th surv ey r e sul t s showed no signif icant c hang es in on-the-job behavior six weeks after training. In contrast the "immediate reaction" ratings were positive, implying the training program was a "success." Conclusions w ere drawn concerning the validity of methods compared.