Data covering a ten-year period in Milwaukee, Wisconsin, are used to estimate the effect of race and other variables on sentences given to persons charged with armed robbery and burglary. Longitudinal data, coupled with interviews of participants, are used to overcome problems of sample selection bias usually associated with cross-sectional studies of the determinants of sentencing. The research shows that race had a clear effect on both the decision to imprison and the length of prison terms in the earliest period (1967-1968) but not in the two later periods (1971-1972 and 1976-1977). The racial neutrality of sentencing in the later periods appears to be the result of changes in the composition of the judiciary, a greater bureaucratization of the prosecutorial and defense bar, and the rise of decision rules that reduce the effect of judicial ideology on outcomes.