Length of postnatal hospital stay has declined dramatically since the 1970s, with ongoing controversy about potential harmful effects. Three population-based surveys of recent mothers conducted in the State of Victoria, Australia have been analysed to assess the impact of shorter length of stay on breast feeding and women's psychological well-being. Women giving birth in Victoria, Australia in 1 week in 1989, 2 weeks in 1993 and 2 weeks in 1999, excluding those who had a stillbirth or neonatal death, were mailed surveys 5-8 months postpartum. Adjusted response fractions were: 71.4% in 1989 (n = 790), 62.5% in 1994 (n = 1313), and 67% in 2000 (n = 1616). Participants were representative in terms of method of birth, parity and infant birthweight. Younger women, single women and women of non-English-speaking background (born outside Australia) were under-represented. The primary outcome measures were infant feeding at 6 weeks postpartum and maternal depression at 5-8 months postpartum (Edinburgh Postnatal Depression Scale > or = 13). There was no significant association between length of stay (1-2 days vs. > or = 5 days, 3-4 days vs. > or = 5 days) and primary outcome measures in univariable analyses of the 1989 Survey, or multivariable analyses of the 1994 and 2000 Surveys adjusting for relevant social and obstetric factors. For stays of 3-4 days, the adjusted odds ratio for formula feeding at 6 weeks was 1.35 [95% CI 0.9, 1.9] in 1994 and 1.22 [95% CI 0.9, 1.7] in 2000. The confidence intervals are compatible with a very small reduction or a large increase in formula feeding, neither reaching statistical significance. For depressive symptoms at 5-7 months postpartum (EPDS score > or = 13), the adjusted odds ratio for women staying 3-4 days was 0.96 [95% CI 0.7, 1.4] in 1994 and 0.90 [95% CI 0.6, 1.3] in 2000. These confidence intervals are compatible with a 30-40% reduction or a 30-40% increase in odds of depressive symptoms. Based on these findings shorter length of stay does not appear to have an adverse impact on breast feeding or women's emotional well-being. Testing early discharge policies in well-designed randomised trials remains a priority for developing stronger evidence to inform practice.