THE SERIES OF FEDERAL CRIMINAL SENTENCING IN THE UNITED STATES WAS BORN OUT OF THE BELIEF THAT A COMPREHENSIVE ASSESSMENT OF EXPERIENCE IS A NECESSARY PRELUDE TO PRODUCTIVE REFORM. USING A COMBINATION OF STEP-WISE MULTIPLE REGRESSION ANALYSIS AND PREDICTIVE ATTRIBUTE ANALYSIS, IT ATTEMPTS TO ASCERTAIN THE PRECISION WITH WHICH ONE CAN STATISTICALLY 'EXPLAIN' VARIATIONS IN CRIMINAL SENTENCES IMPOSED AGAINST OFFENDERS CONVICTED IN 1971 OF EIGHT FEDERAL OFFENSES (ROBBERY, AUTO THEFT, LARCENY, COUNTERFEITING, EMBEZZLEMENT, VIOLATIONS OF THE SELECTIVE SERVICE ACT, MARIJUANA VIOLATIONS, AND NARCOTIC DRUG OFFENSES). AT THE SAME TIME, IT IDENTIFIES, FROM A VARIETY OF OFFENDER, OFFENSE, PROCESS, AND COURT-RELATED FACTORS, THOSE VARIABLES WHICH APPEAR TO EXPECT THE GREATEST EFFECT ON SENTENCE OUTCOME, ASSESSING THE RELATIVE INDEPENDENT CONTRIBUTION OF EACH. OVER 50 PERCENT OF THE TOTAL VARIATION IN SENTENCES IMPOSED IN 1971 COULD BE EXPLANED USING MULTIPLE REGRESSION ANALYSIS. THE FACTORS WHICH APPEARED MOST STRONGLY RELATED TO SENTENCE OUTCOME AT THE AGGREGATE NATIONAL LEVEL WERE THE CONVICTION OFFENSE, THE OFFENDER'S PRIOR RECORD, AND THE METHOD OF CONVICTION. PREDICTION LEVELS DROPPED MARKEDLY WHEN OFFENSES WERE ANALYZED INDIVIDUALLY, RANGING FROM LESS THAN TEN PERCENT FOR EMBEZZLEMENT TO SLIGHTLY MORE THAN FORTY PERCENT FOR NARCOTICS OFFENSES. BRIEF EXPLANATIONS OF THE 23 INDEPENDENT VARIABLES USED IN THE ANALYSIS ARE APPENDED. FOR THE OTHER MONOGRAPHS IN THE SERIES, SEE NCJ-36883, NCJ-36885, AND 86. (AUTHOR ABSTRACT MODIFIED) (SNI ABSTRACT)