Two samples provided the study data. The Chicago sample consisted of 510 cases brought to the Sentencing Council of the United States District Court for the Northern District of Illinois between January 1, 1972, and June 30, 1973. The New York sample was composed of 471 cases brought to the Sentencing Panel of the United States District Court for the Eastern District of New York during 1973. Measures of offender and offense characteristics were obtained from the presentence reports of each case. Measures considered include offense characteristics, criminal record, offender background, cooperation with the court, and conflict between aggravating and mitigating circumstances. The primary sentencing decision to be made was whether to incarcerate the offender. To obtain a continuous measure of sentence disparity, the sentence recommended by each judge was given a point value, and a measure of disparity among judges in a case was then computed. Conflict appeared to be the strongest source of disparity in Chicago, and female offenders evoked less disparity than males. Disparity was greater with more educated offenders in New York, while disparity was higher in Chicago for offenders with health problems and drug addiction. Disparity occurs when judges differ in their overall severity levels, when systematic differences exist between judges in weighing particular case characteristics, and when both aggravating and mitigating factors are present. Policy considerations mandate the use of harnessed judicial discretion informed by a serious examination of the identified sources of disparity. Ten tables, 21 footnotes, and 21 references are provided.