ABSTRACT. In this article, we focus on the early development of autism studied through family home movies. We review all investigations published in English that met specific methodological standards, including the use of comparison samples, coding blind to group membership, and adequate levels of interrater reliability. After discussing in detail the pros and cons of the home-movie methodology, we review the results of all empirical studies conducted to date. We then present a summary of the features found consistently across studies that differentiate autism from typical development and mental retardation in the first 2 years of life. How family home movies can contribute to our understanding of the regression phenomenon is also addressed. Finally, the results are interpreted from both a theoretical and clinical point of view.