Arrest rates per capita for possession of marijuana have increased threefold over the last 20 years and now constitute the largest single arrest offense category. Despite the increase in arrest numbers, rates of use have remained stable during much of the same period. This article presents the first estimates of the arrest probabilities for marijuana, conditional on use in the previous 12 months; this is an appropriate measure of the intensity of enforcement against users. We analyze differences by age, race, and gender from 1982 to 2008. The probabilities of arrest for a marijuana user were similar across age and race categories until 1991. By 2006, that had changed sharply. Arrest rates among current marijuana users are disproportionately high for adolescents, Blacks, and males. The rate has varied between 0.8% and 1.8% across years; the rate per incident of use has ranged between about 1/3,000 and 1/6,000. There is no compelling account of why marijuana arrest probabilities have increased nationally or why the focus has been on youth, minorities, and males but the disproportionate increase for young Black males raises issues of disparate impact. (Contains 17 notes, 4 tables and 10 figures.)