The participants were the 2,748 individuals 20-to 64-years-old who received services from community mental health programs for people with severe and persistent mental illness in Vermont during 1996. The research used Probabilistic Population Estimation to measure the number of people shared between anonymous computerized data sets. The data sets were the Quarterly Service Report Data Base maintained by the State mental health authority and the database of all criminal charges filed during 1997 as recorded by the Vermont Center for Justice Research. Results revealed that 6.8 percent of the people who had received the community services for severe and persistent mental illness during 1996 were charged with a crime during 1997. This figure included 6.3 percent of the females and 7.5 percent of the males. Results indicated that members of the treatment group were 4.4 times as likely to be charged with a crime as were members of the State’s general population. The elevated risk was substantially greater for women and for older treatment recipients. This elevated risk could result either from behavioral differences or from law enforcement practices. Findings indicated the need for service coordination, cross training, and criminal justice diversion programs for people with severe mental illness and for further research that makes comparisons in more detail. The analysis also concluded that the methodology of the study can be useful both to policy researchers and to mental health and criminal justice program administrators. Tables, figure, and 30 references (Author abstract modified)