THE APPROACH OF THIS STUDY IS RELATED TO RECOMMENDATIONS ON PRETRIAL DETENTION BY THE NATIONAL ADVISORY COMMISSION ON CRIMINAL JUSTICE STANDARDS AND GOALS. THE COMMISSION RECOMMENDED USE OF THE LEAST INTERVENTIONARY FORM OF PRETRIAL RELEASE NECESSARY TO ASSURE COURT APPEARANCE. HOWEVER, THERE ARE ALSO QUESTIONS OF FURTHER CRIMINAL ACTS BY PERSONS ON PRETRIAL RELEASE AND OF CONTROVERSIAL PROPOSALS FOR PREVENTIVE DETENTION AND CIVIL COMMITMENT. THE DETENTION ALTERNATIVES DISCUSSED COVER THREE GROUPINGS: USE OF SUMMONS IN LIEU OF ARREST WARRANT; CITATION IN LIEU OF ARREST AND DETENTION; AND PRETRIAL RELEASE AFTER BOOKING. A NORMAL INCIDENT OF ARREST HAS HISTORICALLY INCLUDED DETENTION TO INSURE THE APPEARANCE OF THE ARRESTEE AT TRIAL. THIS APPROACH IS STILL WIDELY CONSIDERED TO BE NORMATIVE, ALTHOUGH THERE HAVE BEEN SOME ALTERNATIVES TO PHYSICAL ARREST AND DETENTION UTILIZED OVER THE YEARS. IN THE 1960'S, THE MANHATTAN BAIL PROJECT WAS ESTABLISHED TO TEST THE RELATIONSHIP OF NONMONETARY BOND RELEASE AND THE LIKELIHOOD OF APPEARANCE AT TRIAL. CERTAIN BACKGROUND CHARACTERISTICS BASED ON RESIDENCY, LENGTH OF EMPLOYMENT, AND NATURE OF THE OFFENSE WERE USED TO ESTIMATE THE STRENGTH OF THE ARRESTEE'S TIES WITH THE COMMUNITY AS AN INDICATOR OF RELIABILITY. THE RESULTING PROGRAM WAS TERMED RELEASE ON RECOGNIZANCE (ROR). AN OUTGROWTH OF THE MANHATTAN EXPERIMENT WAS THE PROLIFERATION OF SIMILAR ROR PROGRAMS AROUND THE COUNTRY. ON THE BASIS OF A REVIEW OF SUCH PROGRAMS CERTAIN ASSUMPTIONS CAN BE ADVANCE: (1) THE EARLIER A DECISION NOT TO DETAIN AN ACCUSED PERSON CAN BE MADE AND IMPLEMENTED, THE LESS COST TO THE TAXPAYER AND THE ARRESTEE; (2) A JURISDICTION COULD SET A GOAL OF ALMOST ANY LEVEL OF RELEASE USE WITHOUT EXCEEDING RATES TO BE FOUND ELSEWHERE IN THE COUNTRY, BECAUSE THERE ARE NO EFFECTIVE UPPER OR LOWER LIMITS; (3) IT IS NOT POSSIBLE TO PREDICT A SPECIFIC FAILURE RATE FOR A GIVEN LEVEL OF PRETRIAL RELEASE; (4) THE AVERAGE LENGTH OF TIME IN PRETRIAL RELEASE STATUS IS A MAJOR FACTOR IN FAILURE RATES -- WILLFUL FAILURE TO APPEAR OR REARREST ARE MORE LIKELY TO OCCUR THE LONGER A FINAL DISPOSITION IS DELAYED; AND (5) CHARACTERISTICS AND CIRCUMSTANCES OF THE DEFENDANT APPEAR TO BE PREDICTIVE OF FAILURE OR SUCCESS, BUT NOT STRONGLY SO, NOR ANY ONE FACTOR TAKEN ALONE. GRAPHIC DATA ARE INCLUDED. (KBL)