Studies on sexual minorities’ social, demographic, and economic outcomes using survey data have long been hampered by data shortfalls, with most survey data limited by sample size or absence of direct questions on sexual orientation. Despite its increasing availability, administrative data has not been systematically utilized to fill the gaps. This article discusses the strengths and limitations of a novel administrative dataset for research on individuals in same-sex couples. Using the Longitudinal Administrative Databank (LAD), consisting of 20% of Canadian tax filers, we recommend strategies of adjusting for potential biases in key measures; comparing LAD estimates to Canadian Census; and introduce the measure of inferred sexual orientation given the lack of direct measure of sexual orientation in tax data. Administrative tax data has major strengths, including a large sample size, high quality income data of individuals and linked family members, and longitudinal design. However, we spotlight three issues related to the identification of same-sex couples: underreporting, misclassification, and measurement errors. We show that some biases are reduced after implementing adjustment strategies. In recent years, the estimated share of same-sex couples is close to that in the 2016 Canadian Census. We also demonstrate the usefulness of a new measure “inferred sexual orientation” by estimating sexual orientation earnings gaps. This article contributes to research on global sexual minority data landscapes and methodologies of identifying and measuring sexual minority populations in longitudinal administrative tax data.