To communicate health risks associated with short-term changes in air pollution, the US EPA reports local air quality through the Air Quality Index (AQI). However, it remains unclear whether the current regulatory-based, single-pollutant AQI fully represents the actual risks of air pollution-related illness. A revised index using a multi-pollutant approach based on health effects could potentially improve public health by better reflecting the health risks associated with exposure to multiple pollutants. Using time-series analysis, daily AQI values of four criteria pollutants (NO2, O3, PM2.5, and SO2) in Bronx and Queens Counties of New York from 2005 to 2010 were regressed against total respiratory emergency department (ED) visits using a Poisson generalized linear model to generate region-specific coefficients (NO2: 0.011; O3: 0.0027; PM2.5: 0.0022; SO2: 0.0013), which were used to develop a multi-pollutant health-based air quality index. Multi-pollutant index models and single-pollutant EPA AQI values were regressed against total respiratory ED visits from 2011 to 2013 to determine the association of index values with population-level health outcomes. Based on time-series analysis, each pollutant considered was shown to have significant positive associations with respiratory ED visits for at least part of the year and was therefore eligible for potential inclusion in a multi-pollutant index. A log-transformed, multi-pollutant health-based model with NO2, O3, PM2.5, and SO2 was found to have more consistent associations throughout the high-O3 (April–September) (1.03, 95% CI [1.01–1.05]) and low- O3 (October–March) (1.03, 95% CI [1.01–1.05]) seasons with total respiratory ED visits as compared to AQI values. Associations between respiratory ED visits and the AQI were not significant during the high-O3 season (high-O3: 1.00, 95% CI [0.99–1.02]. These results indicate that a single-pollutant index may at times inadequately communicate the full adverse health risks of air pollution. A multi-pollutant index, that was adjusted to ensure a relatively normal distribution of index values, was able to reflect population level health outcomes during the high-O3 season when air pollution mixtures become more complex, while the currently utilized AQI could not represent population-level air pollution health risks during the high-O3 season. The development and validation of a multi-pollutant index for use in the US is something that may merit consideration in future updates to air quality standards under the Clean Air Act. Local jurisdictions may wish to act sooner to improve risk communication of outdoor air pollution.