The sample consisted of every other felony probationer processed in Harris County criminal district courts between May 12 and July 17, 1989 (658 cases). The discriminant and regression analyses included the following independent variables: offense, age, sex, race, number of prior felony convictions, number of prior misdemeanor convictions, previous drug offense status, treatment status, classification risk and needs scores, number of positive drug tests, number of contacts with probation officer, types of drugs for which probationer tested positive, and number of drug tests administered. The dependent variable was "successful" or "unsuccessful" probation. Probation failure consisted of a law violation during probation, probation revocation due to law or technical violations, absconder status, or unsuccessful termination. Successful probationers were tested and seen by officers a significantly greater number of times than unsuccessful probationers. Drug and nondrug offenders had similar law-violation rates, but drug offenders committed a higher number of technical violations. Evidence of drug abuse was related to probation failure. Probation success was enhanced for drug offenders with characteristics predictive of failure. 9 tables and 8 references