Earlier research has used 1980 and 1990 U.S. census data to examine the effectiveness of disproportionate stratification to oversample areas with greater concentrations of a minority population in order to achieve a specified effective sample size for a national survey of that minority. The areas considered were census block and block groups. The effectiveness of this oversampling depends on the degree of the minority's geographical concentration and on the relative cost of the full data collection to the screening cost. This paper updates the earlier findings using 2010 U.S. census data and compares these findings with those based on data from the 1990 U.S. census. The paper extends the application to subnational surveys that are concerned with a single census region or with core-based statistical areas (CBSAs) or non-CBSAs. It also examines the effect of using different cutpoints for defining the density strata. Oversampling two minority populations in a survey and oversampling a minority as part of a survey of the overall population are also discussed.