We conduct a laboratory experiment that tests two fundamental predictions unique to salience theory. If an agent purchases one of two vertically differentiated products, salience theory makes the following two distinct predictions. First, it hypothesizes that a higher expected price level for both products shifts demand toward the more expensive, high-quality product. Second, it predicts that demand for the high-quality product is larger if the p rice level is expectedly high than if it is unexpectedly high. In our experiment, subjects purchased fast or slow Internet access at different price levels. Our results strongly support both predictions of salience theory.