As the empirical evidence accumulates, so does confidence that carefully designed and delivered rehabilitation approaches can reduce risk. Yet little is known about how to rehabilitate some specialized groups, such as high-risk violent offenders: career criminals with an extensive history of violent behavior. Since 1998, New Zealand's Rimutaka Violence Prevention Unit (RPVU) has provided intensive cognitive-behavioral rehabilitation to violent men. In this evaluation, 112 medium- and high-risk prisoners who entered the program after 1998 are case matched to 112 untreated men. Reconviction outcome data over an average of 3.5 years postrelease show that 10% to 12% fewer program completers were reconvicted for violence compared to their untreated controls. High-risk completers also are less likely to be reconvicted for any offense. Those men who opted out of the study are a slightly higher-risk group than those who completed it, but noncompletion does not further increase their risk. Given the lack of program theory, and formidable practical challenges involved in working with such a high-risk group, these results are very promising. (Contains 4 notes, 3 tables, and 1 figure.)