Assessment is a necessary task in all areas of education, but there is no agreement on how to assess the impacts of different approaches to sex education, both on an individual level and on a population level over time. The history of mid-20th Century Family Life Education in the United States illuminates some of the obstacles that have made assessing sex education programmes so difficult: control groups, access to large numbers of research subjects and the means to verify self-reporting are elusive. These persistent challenges have to do with the nature of the subject, which is, in contrast to most subjects, not supposed to be practised at school. Standards of reliability, validity and classroom authenticity, therefore, apply partially at best. We argue that some approaches to sex education are valuable whether or not they are assessable, and that some things that are assessable may not be valuable in the way they are thought to be.