Clinicians in several countries have expressed concerns about possible adverse effects of shortening obstetric length of stay. A population-based survey of 1366 mothers who gave birth in Victoria, Australia, in 1993 was used to investigate social and obstetric characteristics of mothers discharged home 'early', and to assess whether shorter stays were associated with adverse health outcomes, or a lesser degree of satisfaction, or both. Women's views and experiences of length of hospital stay were gathered via a statewide postal survey of women who gave birth in a 2-week period; 62.5% (n = 1336) responded. Assessment of the relationship between length of stay (1-2 days vs. > or = 5 days, and 3-4 days vs. > or = 5 days) and four main outcome measures (infant feeding at 6 weeks, period prevalence of feeding problems, maternal confidence and depression) showed no association between these variables and length of stay after adjusting for other obstetric and social factors in separate regression analyses. For stays of 3-4 days, the adjusted odds ratio (OR) for formula feeding at 6 weeks was 1.35 [95% confidence interval (CI) 0.9-1.9]; for feeding problems OR = 0.87 [0.7-1.2]; for lacking confidence OR = 0.81 [0.6-1.2]; and for depression OR = 0.96 [0.7-1.4]. Large randomised trials of early obstetric discharge are required to resolve continuing uncertainties about the safety, and possible benefits of shorter hospital stays.