Objectives: The adverse effects of hot weather on public health are of increasing concern. A surveillance system using 911 medical dispatch data for the detection of heat-related illness (HRI) could provide new information on the impact of excessive heat on the population. This paper describes how we identified medical dispatch call codes, called "determinants", that could represent HRI events. Methods: Approximately 500 medical dispatch determinants were reviewed in focus groups composed of Emergency Medical Services (EMS) paramedics, dispatchers, physicians, and public health epidemiologists. Each group was asked to select those determinants that might adequately represent HRI. Selections were then assessed empirically using correlations with daily mean temperature over the study period (June 1 - August 31, 2005). Results: The focus groups identified 12 determinant groupings and ranked them according to specificity for HRI. Of these, "Heat/cold exposure" was deemed the most specific. The call determinant groupings with the clearest positive associations with daily mean temperature empirically were "Heat/cold exposure" (Spearman's correlation coefficient (SCC) 0.71, p<0.0001) and "Unknown problem (man down)" (SCC 0.21, pEuro Surveillance (Bulletin Europeen Sur Les Maladies Transmissibles; European Communicable Disease Bulletin)0.04). Within each grouping, the determinant "Unknown status (3rd party caller)" showed significant associations, SCCEuro Surveillance (Bulletin Europeen Sur Les Maladies Transmissibles; European Communicable Disease Bulletin)0.34 (pEuro Surveillance (Bulletin Europeen Sur Les Maladies Transmissibles; European Communicable Disease Bulletin)0.001) and SCCEuro Surveillance (Bulletin Europeen Sur Les Maladies Transmissibles; European Communicable Disease Bulletin)0.22 (pEuro Surveillance (Bulletin Europeen Sur Les Maladies Transmissibles; European Communicable Disease Bulletin)0.03) respectively. Conclusions: Clinically-informed expertise and empirical evidence both contributed to identification of a group of 911 medical dispatch call determinants that plausibly represent HRI events. Once evaluated prospectively, these may be used in public health surveillance to better understand environmental health impacts on human populations and inform targeted public health interventions.