This study examined court records, probation records, and collection office records in four counties in Pennsylvania, which were chosen because they varied along two dimensions: (a) population size and (b) the use of specialized units for the collection of monetary sanctions. From each county, restitution-eligible cases were sampled from both 1994 and 1996 to test the effect of a 1995 statutory change mandating restitution. Multivariate models indicated that restitution was significantly more likely to be ordered for property crimes, for offenses that were more easily quantified, for offenses against businesses, and for offenses after the statutory change. Moreover, restitution was more likely to be imposed and a higher percentage was likely to be paid in counties with smaller populations and in counties in which probation officers handled the collection of economic sanctions than in counties in which they were handled by specialized collection units. (Contains 5 tables and 7 notes.)