To maximise the effectiveness of psychiatric day hospitals it is important to establish which patients benefit most. We tested the hypothesis that day hospital therapists can predict responders. The consultant, key-workers, junior psychiatrist and secretary predicted outcome for 26 patients. These were measured blind using the Brief Psychiatric Rating Scale (BPRS), Global Assessment Scale (GAS), Becks Depression Inventory (BDI) and the Social Functioning Questionnaire (SFQ). There was poor correlation generally between staff predictions and patient progress as measured by the standardised instruments. The only significant correlation was the consultant's prediction with the BPRS. We suggest this is consistent with the consultant's experience and training in phenomenology. We conclude that consultants should be fully involved in day hospital assessments.