Objective: To examine the influence of the nurse, the type of patient presentation and the level of hospital service on consistency of triage using the Australasian Triage Scale. Methods: A secondary analysis of survey data was conducted. The main study was undertaken to measure the reliability of 237 scenarios for inclusion in a national training programme. Nurses were recruited from a quota sample of Australian ED according to peer group. Analysis was performed to determine concordance: the percentage of responses in the modal triage category. Analysis of variance (anova) and Pearson correlations were used to investigate associations between the explanatory variables and concordance. Results: A total of 42/50 (84%) participants returned questionnaires, providing 9946 scenario responses for analysis. Significant differences in concordance were observed by variables describing the type of patient presentation and level of urgency. Mean scores for the comparison group (adult pain; 70.7%) were higher than the groups involving a mental health or pregnancy presentations (61.4%; P ≤ 0.001; 65.0%; P = 0.02). Modal responses at the extreme ends of the scale were higher than in the middle categories (P ≤ 0.001). There was a significant main effect on concordance by type of service according to peer group (P = 0.03). Of the nine variables that described nurse characteristics, age was the only factor to influence the outcome (P = 0.05). Conclusion: We identified significant problems with the consistency of triage for mental health and pregnancy presentations. Further research is needed to improve the guidelines on the implementation of the Australasian Triage Scale for these populations.