This paper proposes a new approach to jointly model the trading process and the revisions of market quotes. This method accommodates asymmetries in the dynamics of ask and bid quotes after trade related shocks. The empirical spec ification is a vector error correction (VEC) model for ask and bid quotes, with the spread as the co integrating vector, and with an endogenous trading process. This model extends the vector autoregressive (VAR) model introduced by Hasbrouck (Hasbrouck J (1991) Measuring the information content of stock trades. J Finance 46:179 207). We provide evidence against several symmetry assumptions, very familiar among microstructure models. We report asymmetric adjustments of ask and bid prices to trade related shocks, and asymmetric impacts of buyer and seller initiated trades. In general, buys are more informative than sells. The likelihood of symmetric quote responses increases with volatility. We show that our findings are robust across different model specifications, time frequencies, and trading periods. Moreover, we find similar asymmetries in markets with different microstructures.