I propose a novel measure of information share, termed tail information share (TIS), which focuses on modeling the tail dependence of price innovations using copulas. I discuss its detailed technical properties, including unique identifiability, estimation procedures, and statistical properties. The proposed TIS improves over two commonly used measures by providing meaningful economic rationale and unique identifiability. My simulation studies indicate that TIS can yield more accurate estimates of market-specific contributions to price discovery when tail dependence is present. Additionally, I demonstrate the asymptotic consistency and efficiency of TIS estimators. An empirical illustration is provided using a new dataset of high-frequency crude oil futures.