Unlike many studies of learning and pharmaceuticals, this paper considers credence goods such as vitamins and the role of consumer experience in resolving uncertainty when the user cannot observe the effects of the goods after consumption. The Homescan data justifies variations in the purchases: 45% of households choose different Universal Product Code (UPC) items during subsequent shopping trips than the ones they bought originally. My findings suggest that the probability of choosing Brand 1 increases after a positive experience with Brand 1 and declines after a positive experience with Brand 2. This is based on the assumption that the consumer has had a positive experience about the product if she bought it with a current purchase and three periods back. In a structural model I intend to relax this assumption and compare the endogenous speed of learning about vitamins with the speed of learning about drugs.