There are a growing number of large-scale initiatives for twin registers of children. The Australian Twin ADHD Project (ATAP) is used to illustrate two key limitations which may arise with such studies, namely (1) the importance of including or possibly excluding families in which one or both twins have significant developmental disability, and (2) the selective failure to recruit and/or the selective attrition of families in which parents and children share behavioral difficulties. Initially ATAP excluded 1 in 6 of families whose twins were enrolled in the volunteer-based Australian Twin Registry (ATR), and as more children with significant problems were identified, these families were sequentially excluded. With longitudinal data over ten years, two points about retention were identified: the difficulty of retaining the twins in late adolescence, and the loss of the families whose twins had more ADHD symptoms. We discuss strategies for limiting the loss of families and for ensuring comparability of data across registers with similar interests but different methods of recruitment and exclusion.