Objectives: To provide an objective assessment on callers' compliance with NHS Direct advice to attend an accident and emergency (A&E) department. Methods: A representative three week period in May 2000 was investigated. During this period there were no health scares, major health campaigns, or bank holidays that may have affected the call rate. NHS Direct callers who were advised to attend A&E were identified. Data from the four A&E departments for the same three week period and two additional days were searched and matched to NHS Direct data by surname, date of birth, and post code. This process created three groups: (1) callers triaged to A&E who attended, (2) callers triaged to A&E who did not attend, (3) callers with different triage outcome who attended A&E. The age, sex, relationship of caller, time of call, and distance to nearest A&E were compared for groups (1) and (2). Results: Just less than two third of callers triaged to A&E attended with the same presenting complaint. There were no statistically significant differences between group (1) and (2) in terms of age, sex, relationship of caller, time of call, and distance to A&E. A small number of callers (2.4%) were identified as being given other advice and attending A&E for the same presenting complaint. This group took significantly longer to attend A&E than group (1) (χ 2 =139.01, df=7, p<0.001). Conclusions: Assessing levels of compliance is difficult. These findings suggest that NHS Direct may have comparatively high levels of compliance compared with other similar services. However, using the single triage outcome as the means of identifying the advice given may oversimplify the range of possible advice given. The delay in attending A&E for the group of callers who were given other advice may indicate they had tried other actions. Further larger studies are needed to assess the appropriateness of referrals through investigation of clinical outcomes.