A review of the literature on research with similar objectives concludes that although the studies agree on substantive results, they are uneven in their degree of methodological sophistication and strength. Typical technical flaws in the research are failures to use appropriate denominators in the calculation of arson rates, to consider or adjust for irregularities in arson rates, to weigh observations in cross-sectional analyses, and to use multivariate statistical methods and correct for ill-conditioned or multi-collinear data. This study aimed to remedy these methodological flaws. The primary data sources used were the 1980 Census of Massachusetts, Summary Count 3A and 3B; and the Massachusetts Fire Incident Reporting System (MFIRS) tapes of 1983-85. The census file was used to construct a wide range of demographic and socioeconomic variables for the analysis of arson rates. The MFIRS file was aggregated by zip code and census tract to obtain counts of both residential and vehicular fire and arson. The findings confirmed the important roles of poverty and urbanness as factors in arson rates, along with the relatively unimportant role of race. In the absence of major social change in American cities, the most effective strategies for countering the arson problem combine enforcement and legislative initiatives. Examples of effective measures in these areas are provided. 13 tables, 31 references, and appended figures