Over the past decade, the U.S. Census Bureau worked to improve measurement of same-sex couple households in its demographic surveys and the decennial census. This paper presents results from recent testing of the revised relationship question. We have found four ways to greatly improve the quality of the estimate of same-sex married couple households: additional relationship categories; the move to internet response; the addition of a check for consistency between the reports of sex and relationship in data collection via electronic instruments; and use in data processing of what is essentially an implied sex value reported for coupled households via the relationship question. We conclude with a discussion of next steps in the testing and implementation of the revised relationship question on Census Bureau surveys and the census.