We describe a prospective cross-sectional survey over a 12-month period in the principal maternity hospital of Kathmandu, Nepal, where over 50% of the local population deliver. The study aim was to estimate the contribution of birth asphyxia to perinatal mortality in this setting. During 1995, there were 14 371 livebirths and 400 stillbirths, a total stillbirth rate of 27 per 1000 total births. The fresh term (2000 g or more) stillbirth rate was 8.5 per 1000 total births [95% CI 7.1, 10.1]. Ninety-two cases of neonatal encephalopathy (NE) affecting term infants were detected (excluding those due to congenital malformations, hypoglycaemia and early neonatal sepsis). The birth prevalence of NE was 6.4 per 1000 livebirths [95% CI 5.2, 7.8]. There was evidence of intrapartum compromise in 63 (68%) of the cases of NE and 65 (76%) of the stillbirths, but only in 12 (12%) of controls. The cause-specific early neonatal mortality rate for NE was 2.1 per 1000 livebirths [95% CI 1.4, 3.0]. Combining the NE deaths and fresh stillbirths gives an upper estimate for term birth asphyxia perinatal mortality rate of 10.8 per 1000 total births [95% CI 9.2, 12.6], 24% of all perinatal deaths before hospital discharge. This study suggests that birth asphyxia remains an important cause of perinatal mortality in developing countries. The paper discusses the pros and cons of different strategies to reduce birth asphyxia in low-income countries.