The following paper is a useful example of the kind of epidemiological data gathering that is essential before effective countermeasures can be devised or deployed for maximal cost benefit. Regardless of the dedication and the methodological sophistication of the investigator, such studies cannot produce widely generalizable findings if they are carried out on patients whose socioeconomic characteristics cannot be precisely defined and whose differences from the total population at risk cannot therefore be specified. Differences in the patterns of utilization of medical services by various socioeconomic levels have long been recognized. Such differences may be less pronounced with respect to childhood accidents than with respect to prenatal care. But, as the author is careful to point out, they are sufficient to prevent generalization—especially since burn accidents involve both environmental hazard and patterns of child-rearing, both of which differ widely with socioeconomic status. The solution of this problem is not a simple one. It involves longitudinal studies, a very high level of detailed case reporting, and continuous monitoring to ensure that the population studied is a representative sample of the population at risk. The author's conclusions about the influence of poverty-related social conditions on the incidence of burns document further the conclusions of investigators who have studied other types of childhood injuries.