Dangerousness in criminal offenders can be reliably diagnosed and effectively treated with a recidivism rate of 6.1 per cent. We define dangerousness as a potential for inflicting serious bodily harm on another.This is a report of a ten-year study involving 592 male convict ed offenders. Most of the crimes that brought these offenders to our notice were sex offenses. Several were compounded by ex treme violence including murder, manslaughier, assault with intent to kill, and assault with a dangerous weapon. The staff's initial diagnosis indicated that 304 of these persons were not dangerous, and they were released into the community after completing their sentences. Twenty-six (8.6 per cent) subse quently committed serious assaultive (dangerous) crimes.The courts concurred in our diagnosis of dangerous in 226 cases and committed these offenders to our special "treatment" facility for an indeterminate period of one day to life. Following treatment for an average period of forty-three months, eighty- two patients were discharged on recommendation of the clinical staff. Of these, five (6.1 per cent) subsequently committed serious assaultive crimes, including one murder.Forty-nine of the originally committed patients were released by court order against the advice of the clinical staff. Of these, seventeen (34.7 per cent) subsequently committed serious assaul tive crimes, including two murders.Criteria of dangerousness and guidelines for its prediction were elaborated. No tests or psychiatric examinations can depend ably predict a probability of dangerous behavior in the absence of an actual history of a seriously violent assault on another person. The potential for dangerous behavior is relative and covers a wide spectrum, from the mildly dangerous to the ex tremely dangerous. We postulate that those elements of the personality which cause dangerous behavior are common to all classes of assaultive behavior and that there is no basic difference between the man who kills to rape and the one who kills to rob. Treatment is primarily psychiatric. We consider claims for the efficacy of pharmacological and physiological measures to be unproven and even specious. Treatment involves individualiza tion and socialization: the identification and dissipation of an tisocial predatory elements of the personality and the mobiliza tion and reinforcement of socially conformative elements of the personality.