A 12-month prospective study was undertaken to observe current practice and to determine if a Glasgow Coma Scale (GCS) of 8 or less on admission is a useful parameter to predict the need for airway protection in poisoning. For the period of September 1988 to August 1989, there were 414 admissions for poisoning with 3 fatalities (0.7%). A total of 41 patients (10%) were assessed by attending physicians as requiring airway protection. An initial GCS of 8 or less had a sensitivity of 90% and specificity of 95% for predicting the need for intubation. Logistic regression analysis showed an odds ratio of 0.48 for the relationship between GCS and intubation, significant at P < 0.001. In addition, the absence of a gag reflex on admission gave a sensitivity of 70% and specificity of 100% for predicting the need for airway protection. Further analysis showed that the presence or absence of a gag reflex added nothing to the GCS for the prediction of probability for intubation. In conclusion, an initial GCS of 8 or less was found to be a useful guideline for intubation. However, it should be used in conjunction with the clinical context.