A longitudinal study should be undertaken only when there is a clear reason to justify this and a reasonable chance of achieving the objectives. A multidisciplinary approach is often required, and continuity of staff and long-term collaboration with colleagues are important for their success. A longitudinal approach is required to estimate the prevalence of an infection in pregnancy and the rate of vertical transmission and to clarify the natural history in infected children. An example of cytomegalovirus infection is given; the parallels with HIV are obvious.