Although the study by Tarrier et al (Reference Tarrier, Lewis and Haddock2004) appears to be methodologically more rigorous than the similar study of cognitive–behavioural therapy (CBT) in post-acute schizophrenia by Turkington et al (Reference Turkington, Kingdon and Turner2002), I am a little confused by the authors’ conclusions. After clearly demonstrating no superior effect for CBT over supportive counselling on measures of symptom reduction and relapse rates, the authors conclude their paper by stating that they ‘suggest that the optimum psychosocial management of early schizophrenia would include a combination of CBT and family intervention’. Would it be rude to suggest that the authors take into account their own findings before making such a statement? It is also more than a little irritating that the authors refer to their sample as being diagnosed with ‘early schizophrenia’ throughout the paper. Examining the inclusion criteria for this study shows that patients were included who had schizophreniform disorders, delusional disorders and unspecified psychoses. Regarding the criterion for ‘early’, most clinicians would define the duration of schizophrenia from symptom onset to commencement of treatment (hence the concept of the duration of untreated psychosis). Unless the authors specify illness duration, the criterion of ‘early’ cannot be asserted. Thus, this appears to be a study of patients within 2 years of their first episode of non-affective psychoses and not those with early schizophrenia per se.