The current work explores the direct and interaction effects of age, race, ethnicity, and gender disparity on sentence lengths, considering differences between jail and prison sanctions. The liberation hypothesis suggests that increased judicial discretion in cases involving less serious crimes results in greater extralegal disparity; however, this prediction may not be consistent with how sentencing guidelines structure discretion. Drawing on 7 years of data (2004-2010) from the Pennsylvania Commission on Sentencing, we employ quantile regression models to account for the structure of sentencing guidelines, while examining variation in jail and prison sentence length outcomes. Results indicate that punishment severity varies across age, race, ethnicity, and gender subgroups, but not in ways that offer support for the liberation hypothesis or fit with opportunities to exercise discretion under sentencing guidelines. Substantive, methodological, and policy implications are discussed.