The basic epidemiological issues have not changed fundamentally in 30 years but emphases are different. Clarity of concept and definition is essential; we need several definitions for different purposes, and should distinguish global criteria defining groups in some way specifically 'retarded' and partial criteria defining groups not exclusive to retardation. Of global definitions, we can distinguish Intellectual Impairment based on IQ, Learning Disability based on educational criteria, and Mental Handicap or Retardation based on service or administrative criteria. The first and second may be co-terminous for children. The first and third are usually co-terminous below a certain IQ level, conventionally 50, as SII and SMH/R. This is not so for higher ability groups: MII and MMR are conceived and defined differently, and suit different research purposes. There may be much unknown need because there have been so few studies of total IQ defined groups above 50. Although organic and psychological factors are very important, the study of MMR needs to recognize its primarily social nature, reflecting determinants of selection into MMR status in legal, organizational and professional structures, activities and attitudes characteristic of particular communities and cultures. If the primary focus is on aetiological factors, natural history, and preventive possibilities, study group should be aetiologically defined and preferably not limited to MR. The same applies to specific impairments, disabilities, diseases, behaviours and disadvantages. The services are also susceptible to epidemiological approaches, descriptive, analytic, interventionist and evaluative, but little is yet available on agency structure and function, professional activities, attitudes and training, legal contexts, and financial constraints. Rigorous outcome studies would greatly benefit rapidly developing services. We know quite a lot about the distribution and associations of SII/SMR, though more is needed, especially to clarify issues of social class distribution. There is a dearth of good research in MII or MMR, and several large scale new studies could be justified. Work on the epidemiology of aetiological entities has progressed, though there are many gaps. Because of their rarity, we should be looking to bring together all the data on aetiological diagnosis now accumulating in good paediatric units throughout the UK.