Introduction: There is substantial evidence that patients with delusions exhibit a reasoning bias - known as the "jumping to conclusions" (JTC) bias - which leads them to accept hypotheses as correct on the basis of less evidence than controls. We address three questions concerning the JTC bias that require clarification. Firstly, what is the best measure of the JTC bias? Second, is the JTC bias correlated specifically with delusions, or only with the symptomatology of schizophrenia? And third, is the bias enhanced by emotionally salient material? Methods To address these questions, we conducted a series of meta-analyses of studies that used the Beads task to compare the probabilistic reasoning styles of individuals with and without delusions. Results We found that only one of four measures of the JTC bias - "draws to decision" -reached significance. The JTC bias exhibited by delusional subjects - as measured by draws to decision - did not appear to be solely an epiphenomenal effect of schizophrenic symptomatology, and was not amplified by emotionally salient material. Conclusions A tendency to gather less evidence in the Beads task is reliably associated with the presence of delusional symptomatology. In contrast, certainty on the task, and responses to contradictory evidence, do not discriminate well between those with and without delusions. The implications for the underlying basis of the JTC bias, and its role in the formation and maintenance of delusions, are discussed.