A random sample of low-risk pregnant women were equally divided into four groups of 63 nulliparae and multiparae each booked for care in a integrated general practice unit (GPU) and a shared-care (consultant) system. Selection criteria included only women who were admitted because they were in spontaneous labour or thought they were. Nulliparous women booked for shared-care came into hospital at a less advanced state of cervical dilatation than those booked for the GPU and spent longer (11 compared with 8 h) in hospital before delivery; the comparable durations in multiparae were 6 and 4 h. Both the first and second stages of labour were longer in the GPU-booked women but they received less pethidine and fewer had epidural analgesia; they received less electronic fetal monitoring, augmentation and forceps delivery, and fetal distress was diagnosed less often. The 1-min Apgar score was less than or equal to 6 in 17.5% of infants of nulliparae booked for the shared-care system compared with 1.6% of those booked for the GPU. The intubation rate of infants of nulliparae was 11% in the shared-care system compared with no intubations in the GPU. These comparisons demonstrate the simplicity and safety of delivery of low-risk women in the GPU as compared with deliveries of similar women in a shared-care (consultant) unit.