Cognitive-behavioural therapy (CBT) is considered to be effective for the symptoms of schizophrenia. However, this view is based mainly on meta-analysis, whose findings can be influenced by failure to consider sources of bias. To conduct a systematic review and meta-analysis of the effectiveness of CBT for schizophrenic symptoms that includes an examination of potential sources of bias. Data were pooled from randomised trials providing end-of-study data on overall, positive and negative symptoms. The moderating effects of randomisation, masking of outcome assessments, incompleteness of outcome data and use of a control intervention were examined. Publication bias was also investigated. Pooled effect sizes were -0.33 (95% CI -0.47 to -0.19) in 34 studies of overall symptoms, -0.25 (95% CI -0.37 to -0.13) in 33 studies of positive symptoms and -0.13 (95% CI -0.25 to -0.01) in 34 studies of negative symptoms. Masking significantly moderated effect size in the meta-analyses of overall symptoms (effect sizes -0.62 (95% CI -0.88 to -0.35) v. -0.15 (95% CI -0.27 to -0.03), P = 0.001) and positive symptoms (effect sizes -0.57 (95% CI -0.76 to -0.39) v. -0.08 (95% CI -0.18 to 0.03), P<0.001). Use of a control intervention did not moderate effect size in any of the analyses. There was no consistent evidence of publication bias across different analyses. Cognitive-behavioural therapy has a therapeutic effect on schizophrenic symptoms in the 'small' range. This reduces further when sources of bias, particularly masking, are controlled for.