Two randomized experiments were conducted, each with samples of nearly 2,000 persons who were released over a 6-month period in 1976 from the State prisons of Georgia and Texas. Under the name of the Transitional Aid Reserach Project (TARP), persons released from the prisons were assigned either to one of four treatment conditions or to one of two control groups. The treatments provided either eligibility for unemployment benefits at several levels or the alternative of job counseling. One of the two control groups was followed in detail while the other was followed only through administrative records. Data on subjects came from prison records, criminal justice records, unemployment insurance files, and interviews covering postincarceration experiences. Results were analyzed according to a payment equation, the number of weeks employed, number of weeks in prison, and number of property and nonproperty arrests. Researchers found several important implications for the relationship between poverty and crime among-exoffenders. It appears that experimentally induced unemployment does increase arrests for both property and nonproperty crimes, modest transfer payments appear to reduce arrests for property and nonproperty crimes, and modest income transfer payments produce substantial work disincentives. Perhaps the most important substantive conclusion from the TARP experiment concerns the ability of a blend of theory from economics and sociology to explain at least some of the sources of variation in criminal behavior. Tables, footnotes, and 45 references are provided.