Objective: To assess the safety of nurses and paramedics offering telephone assessment, triage, and advice as an alternative to immediate ambulance despatch for emergency ambulance service callers classified by lay call takers as presenting with "non-serious" problems (category C calls). Design: Data for this study were collected as part of a pragmatic randomised controlled trial reported elsewhere. The intervention arm of the trial comprised nurse or paramedic telephone consultation using a computerised decision support system to assess, triage, and advise patients whose calls to the emergency ambulance service had been classified as "non-serious" by call takers applying standard priority despatch criteria. A multidisciplinary expert clinical panel reviewed data from ambulance service, accident and emergency department, hospital inpatient and general practice records, and call transcripts for patients triaged by nurses and paramedics into categories that indicated that despatch of an emergency ambulance was unnecessary. All cases for which one or more members of the panel rated that an emergency ambulance should have been despatched were re-reviewed by the entire panel for an assessment of the "life risk" that might have resulted. Setting: Ambulance services in London and the West Midlands, UK. Study population: Of 635 category C patients assessed by nurses and paramedics, 330 (52%) cases that had been triaged as not requiring an emergency ambulance were identified. Main outcome measures: Assessment of safety of triage decisions. Results: Sufficient data were available from the routine clinical records of 239 (72%) subjects to allow review by the specialist panel. For 231 (96.7%) sets of case notes reviewed, the majority of the panel concurred with the nurses' or paramedics' triage decision. Following secondary review of the records of the remaining eight patients, only two were rated by the majority as having required an emergency ambulance within 14 minutes. For neither of these did a majority of the panel consider that the patient would have been at "life risk" without an emergency ambulance being immediately despatched. However, the transcripts of these two calls indicated that the correct triage decision had been communicated to the patient, which suggests that the triage decision had been incorrectly entered into the decision support system. Conclusions: Telephone advice may be a safe method of managing many category C callers to 999 ambulance services. A clinical trial of the full implementation of this intervention is needed, large enough to exclude the possibility of rare adverse events.