A multi-site evaluation of batterer intervention systems included a four year follow-up of male batterers and their female partners (initial partners at program intake and identified new partners during the follow-up) recruited from four batterer intervention programs in Denver, Colorado, Pittsburgh, Pennsylvania, Dallas and Houston, Texas, in 1995. Approximately 220 men were consecutively recruited at each of the four programs. The men and their female partners were interviewed at the time of program intake and for every three months over the next four years. Besides an extensive background interview at program intake, the men were also administered the MAST and MCMI-III. Their clinical records were collected that included program attendance and a clinical rating form. Arrest records were also collected for the four year follow-up. The primary objective of the research was to improve prediction of continued violence using a more complex analysis that includes both multiple outcomes and conditional factors, with a secondary objective of exploring additional ways to improve prediction.