The LSI, an objective risk classification instrument, was administered to inmates from three jails. Low-scoring inmates from two of the jails were flagged for placement in correctional halfway houses, and the third jail was blind to LSI scores. The halfway house placement rate was 51 percent for the jails that used LSI scores and 16 percent for the jail using traditional subjective classification procedures. The results suggest that subjective offender assessments run the risk of overclassifying offenders whereas objective risk assessments yield more appropriate classifications. 3 tables and 19 references. (Author abstract)