Despite advances in the assessment and treatment of childhood phobias, little is known about their etiology. Rachman has proposed that phobias are acquired through three different pathways: direct conditioning, modeling or instructions/information. We evaluate the empirical support for Rachman's theory in relation to the origins of childhood phobias. Although we find support for Rachman's theory, a number of methodological and theoretical issues are emphasized. For example, insufficient attention has been given to the reliability and validity of retrospective subject reports on the acquisition of childhood phobias. Also some findings on the origins of childhood fears and phobias are more consistent with a nonassociative account of phobia onset, thus providing an interesting challenge to Rachman's theory.