Summary: This paper uses dynamic optimization to study the optimal advertising of fashion products over time. For fashion products, brand advertising and exclusivity are important sales drivers. Therefore, we propose a dynamic model of the sales of multiple styles of a fashion brand based on these variables. The model is estimated using a particle filter method on data from two fashion categories (handbags and sunglasses) and has good fit and prediction. We also derive explicit analytical solutions of the optimal, closed-loop advertising control and use it to explain advertising and fashion cycles. A managerial prescription is that advertising cycling should be out of phase with sales, for example, trend down when fashion sales is trending up. Thus, advertising flattens the fashion sales cycle over time rather than reinforcing it.