The model attempts to incorporate gains and losses of engaging in crime rather than simply testing the deterrence hypothesis. Findings indicate that crime rates are lower in areas with higher incomes and higher 'costs' of crimes. The certainty and severity of imprisonment are inversely related to crime rates even using data corrected for underreporting. However, the certainty of apprehension can lead to either increases or decreases in crime rates. While the findings thus support the deterrence hypothesis, an important qualification becomes evident. The magnitude of the effect of legitimate opportunities on crime may be larger than previously estimated using uncorrected crime data. A more intuitive specification of the crime supply function using increments of the distributions of returns to work and crime produced results suggesting that the joint risk of engaging in work and crime is a major determinant of crime rates. Low income, high turnover, secondary labor market jobs can provide easy access to criminal work; burglary and larceny are frequently inside, on-the-job crimes. The returns to burglary and larceny are positively correlated with work returns. If the joint risk of engaging in work and crime is higher in the primary labor market, then moving workers out of the secondary labor market into better jobs will reduce these types of thefts. Labor market strategies for reducing crime among ex-offenders should take into consideration the interactions between legitimate and illegitimate opportunities. Tables, 13 references, and an appendix of data sources are included.