DATA WERE GATHERED FROM POLICE RECORDS ON 3,847 OFFENSES, PRIMARILY ASSAULT, UNAUTHORIZED POSSESSION OF ALCOHOL, PETTY ASSAULT, LOITERING, AND OBSTRUCTION OF AN OFFICIAL. CROSS-TABULATION, CORRELATION AND REGRESSION ANALYSIS, AND SCATTER DIAGRAMS WERE USED TO DETERMINE WHAT NEIGHBORHOOD CHARACTERISTICS WERE ASSOCIATED WITH OFFENDER RATES (NUMBER OF POLICE SUSPECTS RESIDING IN AN AREA PER 10,000 POPULATION OVER AGE 14). AREAS WITH HIGH OFFENDER RATES WERE CHARACTERIZED BY UNDER-AVERAGE SOCIOECONOMIC LEVEL; ABOVE-AVERAGE PERCENTAGE OF POPULATION LIVING IN CROWDED QUARTERS, IN CITY-RENTED HOUSING, AND IN RENTAL HOUSING GENERALLY; AND LOW 'FLOOR-AREA RATIO' (TOTAL AREA OF ALL BUILDING FLOORS DIVIDED BY LAND AREA). DEPENDING ON THE PARTICULAR AREA, HIGH OFFENDER RATES WERE ALSO ASSOCIATED WITH HOUSING THAT WAS EITHER VERY OLD OR VERY NEW, AND WITH A LOW PERCENTAGE OF FLOOR SPACE DEVOTED TO RESIDENTIAL PURPOSES. THE REVERSE OF THESE CHARACTERISTICS TENDED TO HOLD TRUE FOR AREAS WITH LOW OFFENDER RATES. PERCENTAGE OF THE POPULATION BELONGING TO THE LOWEST SOCIAL CLASS WAS BY FAR THE BEST PREDICTOR OF HIGH OFFENDER RATES. THE EXPLANATORY POWER OF MANY OTHER VARIABLES DWINDLED OR DISAPPEARED WHEN THE SOCIAL CLASS VARIABLE WAS HELD CONSTANT, ALTHOUGH AGE OF HOUSING HAD A CLEARLY INDEPENDENT EFFECT. THE EXTENT TO WHICH SOCIAL CLASS EXPLAINED OFFENDER RATE VARIED BETWEEN AGE GROUPS OF OFFENDERS (YOUTHS AND ADULTS) AND TYPES OF AREAS (URBAN AND SUBURBAN). FOR INSTANCE, SOCIAL CLASS AND OFFENDER RATE FOR YOUTHS WERE VIRTUALLY UNRELATED IN CITY DISTRICTS WITH MANY NEW AND OLD HOUSES, AND IN SUBURBAN DISTRICTS WITH FEW OLD OR NEW HOUSES. OVERALL, THE CONNECTION BETWEEN SOCIAL CLASS AND OFFENDER RATE WAS STRONGER FOR YOUNG OFFENDERS THAN FOR ADULT OFFENDERS. STATISTICAL TABLES, GRAPHS, METHODOLOGICAL DETAILS, AND A LIST OF REFERENCES ARE PROVIDED. --IN FINNISH, WITH A 17-PAGE ENGLISH SUMMARY. (LKM)