Evidence for an association between mental illness and violence has grown in recent years, leading many to ask why such an association exists. One hypothesis links elevated rates of violence among people with mental illness to a small set of psychotic symptoms--so called threat/control-override (TCO) symptoms. Several studies have supported this hypothesis, but none has examined which of the components, threat or control-override--if either--predominates in explaining violence. To explore this issue we used data from a two-stage epidemiological study (n = 2741) conducted in Israel. Data on TCO symptoms were collected using two methods--fixed-format self-report questions from the first stage and psychiatrists' ratings based on interviews using the Schedule for Affective Disorders and Schizophrenia (SADS) from the second. Results show that both a measure of threat and a measure of control-override are independently associated with violent behaviors. Results also show that neither method--neither fixed-format questions nor psychiatrist rating--predominates in explaining violence. In sum, these results indicate that both the threat and the control-override components of the TCO concept are useful in predicting violent behaviors and that a better measurement of the TCO concept is achieved using a multimethod approach.