In Study 1, the LSI was used to identify potential targets for diversionary intervention. Over two-thirds of the inmates assessed by the LSI as suitable candidates for correctional halfway houses were instead transferred to institutions. In Study 2, LSI scores were used to refer inmates directly to a halfway house review board for possible community placement. The results showed that inmates were transferred to the halfway houses earlier in their sentences, thus increasing the average length of stay in the houses with no increased risk to the community. Despite the use of the LSI to identify potential halfway house candidates, a large proportion of minimum-risk inmates were still transferred to institutions. The results from both studies suggest a tendency to overclassify inmates and that the use of an objective classification instrument can serve to minimize overclassification. In addition, the studies demonstrated predictive validity for the LSI with a representative sample of inmates and improved prediction with increased sampling of the criterion behavior. (Author abstract)