Rather than using population figures, this study used the number of buildings to calculate arson risk. The measurements were transformed for skewness and all observations were weighted proportional to their population counts. Because of the high degree of multicollinearity among the census variables, a factor analytic approach was used to extract the most prominent socioeconomic dimensions to the profile. The results indicate that urbanness and poverty are the two factors most strongly influencing arson rates; the housing quality factor is significant in residential arson but irrelevant to vehicle arson. The race effect was much weaker than in most other types of crimes and age/sex variables were also insignificant. Car arson rates vary widely among cities; rates are generally more difficult to explain across all specifications. As initiatives taken in Massachusetts demonstrate, law enforcement and legislation can make inroads in alleviating arson. 6 tables, 1 figure, and 18 references