Purpose: Although some studies suggest that art therapymay be useful in the treatment of negative symptoms ofschizophrenia, a recent large trial of group art therapyfound no clinical advantage over standard care, but thestudy population was heterogeneous and uptake of theintervention was poor. This study aimed to investigatewhether art therapy was more effective for specific sub-groups of patients.Methods: Secondary analysis of data from a randomisedcontrolled trial of group art therapy as an adjunctivetreatment for schizophrenia (n=140) versus standard carealone (n=137). Positive and Negative Syndrome Scalescores at 12 months were compared between trial arms.Interaction between intervention effect and different sub-groups, including those with more severe negative symp-toms of schizophrenia, and those who expressed apreference for art therapy prior to randomisation, wastested using a linear mixed model.Results: The clinical effectiveness of group art therapy didnot significantly differ between participants with more orless severe negative symptoms [interaction for differencein PANSS=1.7, 95 % CI (-8.6 to 12.1),P=0.741], orbetween those who did and did not express a preference forart therapy [interaction=3.9, 95 % CI (-6.7 to 14.5),P=0.473]. None of the other exploratory subgroupssuggested differences in intervention effect.Conclusions: There was no evidence of greater improve-ment in clinical symptoms of schizophrenia for those withmore severe negative symptoms or those with a preferencefor art therapy. Identification of patients with schizophreniawho may benefit most from group art therapy remainselusive.