The paper reports on a research aimed to evaluate the repeatability of some paediatric data obtained from the parents by using different collection instruments: telephone interview, mail questionnaire, in-person interview. The study included 699 children consecutively born in the Policlinico Universitario "A. Gemelli" from September to November 1983. After 3 years a questionnaire containing questions about measles, anti measles immunization, age at which the baby started to walk, disease delaying the walking, and orthopaedic examinations was sent by mail to 149 families; the other 550 families were searched for by telephone and, if found, asked the same questions. All the contacted families were invited for a paediatric check-up, during which the same data were collected through direct interview. For each question crude agreement and K statistic (which controls for the agreement attributable to chance) were computed with respect to both the phone/direct and mail direct comparisons. 391 families (56%) were traced for the first interview, due to the high number of them which had moved home. Among these 289 (74%) attended the paediatric check-up. On the whole, rather high values of K statistic were observed, ranging from 0.59 (question on diseases delaying the walking, comparison phone/direct) to 0.93 (question on anti-measles immunization, comparison mail/direct). Due to the small sample size, the estimates concerning the comparison mail/direct are rather imprecise. Although its potential is limited by the low response rate, the study brings good evidence that the information considered is not sufficiently reliable when reported retrospectively by parents. However, the keeping of a prospective individual record containing data of medical interest should be encouraged.