AFTER A REVIEW OF THE MAJOR CHANGES IN BAIL REFORM DURING THE 1960'S AND 1970'S, THE VERA SCALE, WHICH IS PROBABLY THE MOST WIDELY USED PREDICTION INSTRUMENT FOR RELEASE ON OWN RECOGNIZANCE, IS EXAMINED. IT WAS FOUND THAT THE SCALE HAS NEVER BEEN WIDELY TESTED EMPIRICALLY AND THAT ITS PREDICTIVE VALUE MAY NOT HOLD TRUE FOR ALL POPULATIONS. IN THIS RESEARCH, THEREFORE, COURT RECORDS FOR ALL FELONY DEFENDANTS RELEASED IN 1975 THROUGH THE RAMSEY COUNTY COURT PRERELEASE INVESTIGATION PROJECT REMAND WERE EXAMINED. THIS GROUP CONSISTED OF 46 RELEASEES WHO FAILED TO APPEAR AND 473 WHO MADE THEIR COURT APPEARANCE ON TIME AND REPRESENTED ABOUT 85 PERCENT OF ALL DEFENDANTS HANDLED THAT YEAR. BOTH MULTIPLE REGRESSION AND DISCRIMINANT ANALYSIS WERE USED. A DETAILED DISCUSSION OF THE SHORTCOMINGS OF MULTIPLE REGRESSION ANALYSIS IN STUDIES OF THIS KIND IS PRESENTED. THE DISCRIMINANT ANALYSIS CLASSIFIED CORRECTLY THOSE WHO DID NOT APPEAR 30 PERCENT OF THE TIME AND THOSE WHO DID APPEAR, 98 PERCENT OF THE TIME. IT IS CONCLUDED THAT FURTHER WORK NEEDS TO BE DONE ON REFINING THE FOLLOWING FAILURE TO APPEAR CLASSIFICATIONS: THE FUGITIVE, THE TECHNICAL FAILURE TO APPEAR, AND THE 'SLOW SHOW.' IT WAS FOUND ALSO THAT THE CATEGORIES ON THE VERA SCALE HAD PREDICTIVE VALUE FOR THIS POPULATION BUT THAT THE WEIGHTINGS WERE NOT VALID. IT IS SUGGESTED THAT EACH AREA DEVELOP ITS OWN PREDICTIVE SCALE BASED ON LOCAL DATA. TABLES PRESENT STUDY DATA. AN EXTENSIVE BIBLIOGRAPHY IS INCLUDED. (GLR)