Since the second world war, despite efforts to control the situation, there has been a substantial increase in mortality and in hospital admissions attributed to poisoning. Some part of this almost certainly represents a real increase in incidence in the general population. This has attracted attention from the viewpoint both of medical care (Dooley, 1962; Goulding and Watkin, 1965) and of pre vention (Kessel, 1965). Not only but also fatalities have increased; and Dooley (1962) showed from the National In-patient Inquiry in the years 1955-9 that hospital admission had increased as much from aspirin as it had from with prescribed medicines. He also pointed out that an upward trend had occurred in other countries. This indicates that, while over prescribing, ready availability of sedative drugs through the National Health Service, and changing reactions to mental stress may each be important, none is the sole answer. Suicidal fatalities are the main problem and the extent to which an impact can be made on these is the crucial preventive question, especially since, with a change of behavioural pattern, the temporal increase in deaths has been partially offset by a reduction in other methods of suicide. Moreover, if suicidal rates are primarily deter mined by cultural and socio-economic factors, a campaign against one method could merely mean replacement by another. Stengel and Cook (1958) have drawn attention to the subtle relationships between successful and suicides and underlined the possible fallacies of considering all episodes as a homo geneous population and of relating increases in the total to increases in deaths. The extent of possible impact by a preventive campaign on suicidal deaths is therefore arguable, but the size of the problem warrants an empirical approach. Sources and Nature of the Data These comprise all admissions for to hospitals in the Cardiff area from 1950-65. Cases of alcoholism and adverse reactions to drugs were excluded. Relevant items were abstracted from e ch case history by a medically-qualified research assistant under the supervision of one of us (J.D.P.G.). In addition to the recording of factual information, cases were divided into four categories: accidental, suicidal (serious attempt), suicidal (gest re), and suicidal (undetermined), the last three assessments being based on the opinion in retrospect of a consultant psychiatrist. In some the attribution of a fatality was clear (e.g., a suicidal n te was left or an accident occurred at work), but in ab ut one-third of fatalities no confident classifica tion could be made. A negligible number of non fat l episodes was doubtful as regards the accidental or suicidal context. Consideration in the present communication is confined to Cardiff residents of whom virtually all at risk would enter the hospitals included in the study. For simplicity of exposition, in view of these categories defined by psychiatric assessment, the older nomenclature of attempted suicide is used in description of the data, but in discussion the more cor ect term of self-administered poisoning intr duced by Kessel (1965) is employed.