A retrospective study of 182 children with birth weights less than or equal to 2,000 gms was undertaken. Its analysis showed that heterogeneity in discharge weights (1,580 to 3,400 gms) corresponded to an homogeneous average gestational age, around 39 weeks. Low discharge weights were neither correlated with the pathology of the first days, nor with the socio-economic status of the family. The majority were children born with intra-uterine growth retardation, who mature more rapidly and who might be given to their families earlier, with lower weights. There was no correlation between discharge weight and a second hospitalization during the first year. A policy of "early" discharge did not seem to have disadvantages, while its advantages are unquestionable.