Factor and correlational analyses have been used to characterize symptom dimensions in schizophrenia, though they have yielded divergent models. This study used meta-analysis of published work to determine the number and composition of symptom dimensions. Principal components analysis of data from 10 empirical studies (pooled n = 896) yielded three factors, 'positive', 'negative' and 'conceptual disorganization'. The findings suggest that a three-factor solution is a relatively stable outcome of studies assessing these symptoms in chronic patients, and that some symptoms (alogia, attentional impairment) are less likely to load uniquely on a single factor.