'Jumping to conclusions' (JTC) on probabilistic reasoning tasks has been shown to be related with delusions in schizophrenia. However, whether JTC is merely correlated with, moderate or mediate delusions is not known. Further, it is unclear how antipsychotics affect JTC and its relationship to delusions. We examined the effect of treatment on JTC in a sample of patients (N=19) who were initiated on treatment and followed. Two versions of the task were used--the 'beads' version of the task and an emotionally salient version. Within two weeks of treatment, we found an increase in the number of trials to decision on the emotionally salient version and a reduction in intensity of psychotic symptoms and delusions (measured by the change on P1 and PANSS-P scores). While, these two measures, or changes in these measures, showed no reliable correlation, the baseline performance on the emotionally salient version of the task helped predict patients who would show improvements in their PANSS-P and global PANSS scores in response to medication. The findings suggest that JTC might moderate the effects of treatment on symptomatology, but it does not mediate the treatment induced reduction in delusional intensity.