The Current Population Survey (CPS), a nationwide household survey conducted by the Bureau of the Census for the Bureau of Labor Statistics, provides official labor force estimates for the working-age noninstitutional population. Variances for the labor force estimators are based on a modified half sample replication method. Since the variances are based on a sample, the variance estimates have a considerable amount of error. One method to remove some of the sampling error in the variances is to group similar labor force items and fit a generalized variance function to the variance estimates. This paper evaluates the current variance estimation procedure and suggests another procedure for generalizing variances for national labor force statistics.