We describe how a modified Gini index serves as an improved method of estimating health care disparities. The method, although general, is applied to an example of birth weight disparities and to their effect on subsequent mortality. The method provides the between-group results obtainable from current methods (i.e. how Hispanics generally fare relative to non-Hispanic Whites) but adds measures of within-group disparities (i.e. which specific Hispanics experience the greatest disparate treatment). Our application to birth weights and receipt of prenatal care, which may provide an upper bound because of omitted variables, shows that the time-of-birth disparities are associated with increased infant mortality within the first year of life. Copyright © 2010 John Wiley & Sons, Ltd.