The reporting of arson is distinguished from that of other index offenses by the requirement that an investigation occur that establishes that a fire has been purposefully set or attempted; the counting of all arsons regardless of their occurrence with other offenses; the detection and reporting of the offense by noncitizens; and the infrequent existence of exclusive police jurisdiction. Because of these unique characteristics, past approaches to assessing the validity of data on index crimes, such as a comparison with victim reports of crime, are not possible or appropriate. In this study UCR data on arson are compared with data obtained through a national survey (n=683) of fire departments. The comparisons indicate that UCR data are significantly lower than the rate of arson reported by local fire departments both overall and across all regions of the country. (Publisher abstract)