In 1994 Link and Stueve identified a number of symptoms--called threat/control-override (TCO) symptoms--that were significantly more than others related to violence. This was confirmed by some, but not all, following studies. The contradictory results could be due to remarkable differences in sample compositions, sources used, and definitions and periods of recorded violence, but they are mainly due to problems defining the TCO symptoms. To reexamine the validity of the TCO concept from an exclusively psychopathological position, we compared in a retrospective design a sample of male offenders with schizophrenia not guilty by reason of insanity (n = 119) with a matched sample of nonoffending schizophrenia patients (n = 105). We could find no significant differences regarding the prevalence of TCO symptoms in the two groups during the course of illness. The only statistically significant discriminating factors were social origin and substance abuse. Yet, taking into account the severity of offenses, TCO symptoms emerged as being associated with severe violence. This effect is primarily attributable to the comparatively unspecific threat symptoms. Control-override, to be seen as more or less typical for schizophrenia, showed no significant association with the severity of violent behavior.