The prevalence of mild mental retardation in 19-year-old survivors of male births during 1944-1947 is derived from military records. The data are singular in that they are national, virtually complete for a total population of more than 400,000 men and include attributes not previously examined. They allow for the simultaneous use of three criteria: education (history of special schooling), psychometric (Raven intelligence test score), and clinical diagnosis (ICD (1948) 325.2, 325.3), which yielded rates per 1000 of 30,58 and 61, respectively. Rates by all three criteria varied in similar fashion with father's occupation and with religious affiliation. Rates by the psychometric and diagnostic criteria were higher for rural- and urban-born, while rates by the special schooling criterion were lower for rural-born. These variations are presumed to indicate deficiencies in rural provisions of special schooling on the one hand, and a substantive increment in "true" prevalence on the other. Marked variations in rates by province are not accounted for by social class and urban/rural birthplace. A rise in rates in the 1947 birth cohort on the psychometric and diagnostic criteria but not on the schooling criterion is attributable to a scoring change in the IQ criterion. Relative risks are estimated for the psychometric criterion of low Raven test score according to father's occupation, urban-rural origin, and religious affiliation. The overlap of identification among individuals designated by one, two, or three of the criteria is examined, and the issue of labeling is explored.