Recently, commentators have championed sentencing by jury as more democratic and better suited to current theories of punishment than sentencing by judge alone. Yet discussions of noncapital jury sentencing either fail to address empirical questions or rely on outdated and incomplete studies. This study is the first to present basic information on the comparative severity and variance of jury and judge sentencing as practiced in the 21st century. We use archival sentencing data from two of the six states that authorize jury sentencing—Arkansas and Virginia. In both states, voluntary sentencing guidelines govern judicial sentencing after bench trial or guilty plea, but juries select sentences after jury trial. Regression analyses evaluated sentences of incarceration for individual offenses, and included as independent variables available offense, offender, and case characteristics associated with sentence severity. For most of the offenses examined, jury sentences after jury trial were both more varied and more severe than sentences selected by judges after bench trial. The findings also suggest that these disparities may be deliberately maintained by judges and prosecutors, and facilitated by state law that restricts the sentencing options and information provided to juries as compared to judges.