The 535 reports of nonoccupational/nonagricultural poisoning with malathion in the Pesticide Incident Monitoring System (PIMS), a data base maintained by the Environmental Protection Agency, were coded and classified to describe the occurrence and to determine risk factors for poisoning. Between 1966 and 1980, 335 (63%) of reported incidents were associated with at least one symptomatic person, with a total of 670 persons exposed. Home exposures accounted for 92% of reports and arose from improper use and labeling problems in 61% and 3% of reports, respectively. Seventeen reports (3%) were associated with commercial extermination at home. One third of the 18 fatalities from malathion were unintentional and, therefore, preventable. The relative risks of fatal outcome from suicidal intent and labeling problems were, respectively 41 (P less than 10[-6]) and 4.8 (P = .09). When data from a previously coded data set on diazinon were added, the odds ratios were 20 and 6.7 (both P less than .0003). This suggests that public health measures aimed at safer use of pesticides outside the workplace are needed and that the PIMS data are a valuable source of epidemiological data on pesticide poisoning.