A new system of care, 'nurse telephone triage' was developed in which patients contacting their general practitioner outside normal surgery hours spoke initially to an experienced and specially trained nurse. The nurse assessed the call, determined the urgency of the call (triage) and made a decision with the caller about how best to manage the situation. The management options included nurse telephone advice, inviting the patient to attend an out of hours surgery, referral to the general practitioner on call, or referral to the accident and emergency. The system was established and integrated within a general practice co-operative and its safety and effectiveness were tested in a block randomised controlled trial over a year. Care provided by a general practice co-operative (the active control) was compared with care provided by the same co-operative augmented by a nurse telephone triage system (the intervention). The key outcome measure was equivalent in the incidence of adverse events, defined as deaths within seven days of a contact with the out of hours service, emergency hospital admissions and attendances to an Accident and Emergency Department within three days of a contact with the out of hours service. Process measures were call management, the characteristics of patients consulting the service, caller experience and patient complaints. The intervention reduced general practitioner workload substantially (telephone advice, home visits and attendances at a primary care emergency centre). Equivalence was observed in the incidence of deaths and emergency hospital admissions. Equivalence in accident and emergency attendance was uncertain. The introduction of a nurse service was an additional cost to a medium sized GP co-operative, but larger co-operatives employing more than one GP on duty could consider reducing the number of general practitioners on duty if a nurse service was introduced. The study has informed the development of other nurse telephone triage systems in the UK including the development of NHS Direct. The trial results do not discourage replication of the tested model.