It is now widely accepted that volatility models have to incorporate theso-called leverage effect in order to to model the dynamics of daily financialreturns.We suggest a new class of multivariate power transformed asymmetricmodels. It includes several functional forms of multivariate GARCH models whichare of great interest in financial modeling and time series literature. Weprovide an explicit necessary and sufficient condition to establish the strictstationarity of the model. We derive the asymptotic properties of thequasi-maximum likelihood estimator of the parameters. These properties areestablished both when the power of the transformation is known or is unknown.The asymptotic results are illustrated by Monte Carlo experiments. Anapplication to real financial data is also proposed.