Tandem mass spectrometry offers the chance to improve newborn screening (NBS) for phenylketonuria and to expand screening programmes at minimal additional costs. So far, however, there are only limited data available on the incidence of a broader range of disorders presently being considered, their natural course, the benefits achievable and potential harm associated with screening. Based on a literature search and experience from the Bavarian extended screening trial, these questions are addressed using medium-chain acyl-CoA dehydrogenase deficiency (MCADD) as an example. The data retrieved are sufficient for estimation of the incidence of MCADD cases identifiable by NBS and for diagnosis following clinical symptoms. Clinically detected cases ascertained by active surveillance in populations with highly developed and freely accessible health care systems consistently amount to only 33% of those identified by NBS. This difference cannot be explained by the difference in the proportion of the homozygous 985A-->G mutation, which accounts for about 50% of cases identified in NBS. Further research is needed to assess the contribution of MCADD to unexplained deaths in infancy. Retrospective cohort studies enrolling at least 500,000 children would allow for a more precise estimate of the natural course of disease in particular with regard to less severe adverse outcomes. The most relevant gap in knowledge concerns the long-term outcome of children identified following symptoms and by newborn screening. Since randomised controlled trials are unlikely to be feasible on this issue, a standardised documentation protocol should be implemented in follow-up studies for cases identified either by high risk screening or newborn screening. A proposal for the content of such observational studies is made.