Treatment discontinuation is common in antipsychotic trials for the treatment of schizophrenia and is often associated with symptom worsening. This study assessed the timing and the magnitude of symptom change associated with treatment discontinuation for any cause, and whether improvement in specific symptom domains was predictive of treatment persistence. This post hoc analysis used pooled data from 1627 patients with schizophrenia spectrum disorders who were enrolled in 4 randomized, double-blind antipsychotic comparator trials of 24 to 28 weeks duration. Mean change in the Positive and Negative Syndrome Scale (PANSS) total scores during the last available visit interval were compared for patients who did and did not complete the trial. Improvements in positive, negative, and depression subscores of the PANSS were used to predict trial persistence. A total of 866 patients (53%) discontinued treatment of any cause during the studies. Patients who discontinued treatment early (weeks 0-4) had no significant improvement in mean PANSS total score during the last visit interval, whereas those who discontinued beyond week 4 had significant worsening of mean PANSS total score in the final visit interval. Improvements in either PANSS positive or depression subscores, but not in the PANSS negative subscore, were strongly predictive of study completion. These findings suggest that early failure to establish treatment response and later symptom worsening are associated with treatment discontinuation for any cause. Early improvement in positive or depressive/anxiety symptoms is especially predictive of treatment persistence.