TARP is based on a smaller study in Baltimore, Md., which indicated that ex-prisoners receiving payments were less likely to be rearrested for property crimes than those who received only job placement or no services of payments of any kind. Data are available on 1,975 Texas parolees and 2,007 Georgia parolees. Released prisoners were assigned to six groups using a stratified random procedure. Four groups consisted of those who received 26 weeks payment and 100-percent tax, 13-weeks payment and 100-percent tax, 13-weeks payment and 25-percent tax, and job placement only. The remaining two groups contained interviewed controls and noninterviewed controls. Nine data files cover prison history; prerelease interviews; 3-month, 6-month, and 12-month interviews; payment summary; and wage, arrest, parole, job market, and workfile data. A card image file contains several decks per case; the OSIRIS dataset is included, along with the variable description list.