Although initially described as an inborn disorder of affective contact, information on autism as it exists in infants has been limited. Delays in diagnosis, lack of information about the condition, and reliance on retrospective research strategies have been problematic. An awareness of the increased risk for siblings is now allowing the development of new, prospective approaches. Consistent with Kanner's original hypothesis, the available information strongly suggests a fundamental difficulty in the earliest social processes, which, in turn, impacts many other areas of development. New approaches to screening have lowered the age of initial diagnosis; this presents new challenges for early intervention. Directions for future research are highlighted.