Since the end of the 1970's, excess mortality among girls, from the end of the neonatal period until the age of 4 years, has been observed in some South Asian countries. Explanatory hypotheses for this situation have in fact noted differences by sex in food allocation and in care during illness. In some North African and sub-Saharan countries in Africa, mortality data suggest the same type of phenomenon, but less reliable statistics and a lack of data analysis by sex on use of health services does not really allow clarification of the problem. The objective of this study was to analyse health seeking behaviour by sex and to identify explanatory factors for any differences found. A cross sectional study of 1560 consultations of under-5 children was carried out in 6 university and regional hospitals in 3 African countries: Algeria (the zones of Aïn Taya and Tigzirt); Togo (the zones of Lomé, Atakpamé and Kara) and the Congo (Brazzaville). Results confirm the existence of discrimination against girls. In ways which vary according to zones, and in comparison to boys, observations of girls show: under-representation among outpatients (29% in Tigzirt, 40% in Kara), especially when they have many siblings, of which some are sisters; longer duration of the development of symptoms before first resort (leading to increased severity of symptoms); less investment in health care; detrimental feeding practices. Higher SES of the father plays a favourable role, especially for girls, and children of both sexes benefit when the mother has a good educational level. The large regional differences demonstrate the complex interaction among explanatory factors: rurality, problems of access to the hospital, low economic level, low social status of women. This research opens the way so that, in Africa, research will no longer be carried out on health seeking behaviour and utilisation of health services without examining separately the situation of boys and girls, and analysing the possible causes of any differences.