The empirical study used a large sample of defendants charged with felonies (burglary, robbery, assault with deadly weapon, receiving stolen property, forgery, and grand theft) in the California Superior Court in 1973. The impact of prior record, criminal status, type of plea, type of defense counsel, offender age, sex, and race, and jurisdictional residence were considered for impact on sentence length. Further analysis considered the impact of jurisdictional variables upon sentence outcome. Some findings were that minority defendants did not appear to be discriminated against in terms of sentence outcomes, and the estimated sentence reductions received by minorities were small. Women received shorter sentences than men, but age did not have a significant impact on outcomes. Jury trial led to a large increase in sentences for all offenses, while bench trial generally led to shorter sentences than those received by defendants who plead guilty initially. The prior record of the defendant had an important impact on plea bargaining dynamics -- either on when to plead guilty or whether to go to trial. Judicial reform, except in certain areas, appeared unlikely to have a major impact upon the elimination of disparity. Legislative reforms were seriously flawed and may actually exacerbate, rather than reduce, the amount of disparity and sentencing variation. A new approach to sentencing reform would use 'sentencing matrices' derived from empirical evidence of disparity. The matrices would mandate much greater formal differentiation of offenders according to their offense, prior record, and aggravating or mitigating circumstances of the crime, and would 'control' the nature and extent of plea bargaining to have minimal impact on fairness.