Goren, Sarty, and Wu (1975) claimed that newborn infants will follow a slowly moving schematic face stimulus with their head and eyes further than they will follow scrambled faces or blank stimuli. Despite the far-reaching theoretical importance of this finding, it has remained controversial and been largely ignored. In Experiment 1 we replicate the basic findings of the study. In Experiment 2 we attempt a second replication in a different maternity hospital, and extend the original findings with evidence suggesting that both the particular configuration of features, and some aspects of the features themselves, are important for preferential tracking in the first hour of life. In Experiment 3 we use a different technique to trace the preferential tracking of faces over the first five months of life. The preferential tracking of faces declines during the second month. The possible causes and consequences of this observation are discussed.