Sequential trade models in the spirit of Easley, Kiefer, O'Hara and Paperman (Journal of Finance, 1996) have recently become a standard application for the empirical analysis of information based trading on asset markets. Assuming that some traders have private information about an asset's true value a number of hypotheses between observable quantities can be established. I introduce an enhanced framework that uses information on trades only for estimation and hypothesis testing of sequential trade models. Thus, I am able to account for a number of possible shortcomings of the original model recently reported by several authors. This econometric framework is applied to a high frequency transaction data set for the common share of IBM traded on the New York Stock Exchange during August 1996.