Quality data has significant ramifications for supply chain members. This study investigates the quality data sharing problem in a two-level supply chain consisting of a manufacturer with private quality data and a retailer. Demand is random and sensitive to price and consumer product quality preference. We develop game models for different channel power structures: manufacturer-led (MS), vertical Nash (VN) and retailer-led (RS). The results indicate that the manufacturer under all three power structures should share data when the quality reaches a specific threshold. The likelihood of the manufacturer to share data is highest under the MS structure, lowest under the RS structure, and falls between the two under the VN structure. The ex ante payoff of the manufacturer always monotonically decreases first and then monotonically increases with the consumer quality preference and the sharing cost. When the sharing cost is high, the manufacturer under the RS structure has higher ex ante payoff. Otherwise, the MS structure is better. When the sharing cost is low, both the supply chain and consumers prefer the VN structure.