We review methodological opportunities and lessons learned in conducting a longitudinal, prospective study of same-sex couples with civil unions, recruited from a population-based sample, who were compared with same-sex couples in their friendship circle who did not have civil unions, and heterosexual married siblings and their spouse. At Time 1 (2002), Vermont was the only U.S. state to provide legal recognition similar to marriage to same-sex couples; couples came from other U.S. states and other countries to obtain a civil union. At Time 2 (2005), only one U.S. state had legalized same-sex marriage, and at Time 3 (2013) about half of U.S. states had legalized same-sex marriage, some within weeks of the onset of the Time 3 study. Opportunities included sampling legalized same-sex relationships from a population; the use of heterosexual married couples and same-sex couples not in legalized relationships as comparison samples from within the same social network; comparisons between sexual minority and heterosexual women and men with and without children; improvements in statistical methods for non-independence of data and missing data; and the use of mixed methodologies. Lessons learned included obtaining funding, locating participants over time as technologies changed, and on-going shifts in marriage laws during the study.