Objectives: To investigate policies on neonatal vitamin K and their implementation. Design: Two phase postal survey. Setting: United Kingdom. Participants: A 10% random sample of midwives registered with the United Kingdom Central Council for nursing, midwifery, and health visiting. Of 3191 midwives in the sample, 2515 (79%) responded to phase one and 2294 (72%) completed questionnaires on their current jobs (November 1998 to May 1999). In phase two, 853 (62%) of 1383 eligible midwives gave details on 2179 of their earliest jobs (start dates before 1990). Results: All the midwives in clinical practice at the time of the survey (2271, 99%) reported that they were working in areas with official policies on neonatal vitamin K. Seven distinct policies were described: intramuscular vitamin K for all babies (1159, 51.0%); intramuscular vitamin K for babies at "high risk," oral for others (470, 20.7%); oral vitamin K for all babies (323, 14.2%); parental choice for all (124, 5.5%); parental choice for all except babies at high risk, (119, 5.2%); intramuscular vitamin K for babies at high risk only (33, 1.5%); oral vitamin K for babies at high risk only (17, 0.7%); and a disparate group of policies including intravenous vitamin K for some babies (26, 1.1%). Previous policies were (and some may still be) open to individual interpretation and were not always followed. Conclusions: Hospital policy is not necessarily a good guide to individual practice. The primary purpose of clinical records is to document patient care, and recording practices reflect this. There is considerable variation in vitamin K policies and midwifery practice in the United Kingdom, and there is no clear consensus on which babies should receive vitamin K intramuscularly.