The prevalence and the natural history of primary speech and language delays were two of four domains covered in a systematic review of the literature related to screening for speech and language delay carried out for the NHS in the UK. The structure and process of the full literature review is introduced and criteria for inclusion in the two domains are specified. The resulting data set gave 16 prevalence estimates generated from 21 publications and 12 natural history studies generated from 18 publications. Results are summarized for six subdivisions of primary speech and language delays: (1) speech and/or language, (2) language only, (3) speech only, (4) expression with comprehension, (5) expression only and (6) comprehension only. Combination of the data suggests that both concurrent and predictive case definition can be problematic. Prediction improves if language is taken independently of speech and if expressive and receptive language are taken together. The results are discussed in terms of the need to develop a model of prevalence based on risk of subsequent difficulties.