In this paper we first present methods and preliminary results of pilot surveys of "serious" mental retardation (IQ less than or equal to 55); the surveys included screening and diagnostic components and were carried out in the less-developed world. Next we discuss two problems raised by these surveys: one is the diagnosis of a case and its clinical dimensions, and the other is the interpretation of prevalence. In the next section we illustrate epidemiological approaches to the analysis of such data, in particular their relevance to prevention. Lastly, we propose that the two-stage survey approach developed in the course of the pilot work can provide a valuable basis for planning and prevention, if certain key conditions can be met.