OBJECTIVE: Over the past decades, there have been significant efforts in the United States to improve pharmaceutical labeling pertaining to pregnancy. The goal of this study was to describe trends in pregnancy labeling at the time of drug approval and over time. STUDY DESIGN: The labeling data of 213 new pharmaceutical approvals between January 2003 and December 2012 were systematically reviewed. Initial approval data and subsequent labeling revisions were evaluated for pregnancy category, source of pregnancy and breast-feeding data, data in labor and delivery, presence of a registry, black-box warnings, and utilization of the new labeling format. RESULTS: The most commonly approved pregnancy category was C (51.6%). Most pharmaceuticals (92.9%) had pregnancy data based on animal studies and 5.2% had human pregnancy data. For breast-feeding, there were no data in 47.9% of labels, animal data in 42.7%, and human data in 4.7%. There were no labor and delivery data in 85.9% of labels. Only 2.8% of medications had human data, with the remainder having animal data. The majority of medications (85%) did not have a pregnancy registry. Of those that had a registry, 68.7% were by therapeutic category, not agent specific. Seven medications had black-box warnings related to teratogenicity. Since the new labeling recommendations, 4.7% of medications incorporated the new format into the labeling, primarily approvals that occurred in 2012. CONCLUSION: Despite significant efforts to improve drug labeling for pregnancy and lactation, there remains a paucity of human data in this understudied population.