Latent variable modeling is a multivariate technique commonly used in the social and behavioral sciences. The models used in such analysis relate all observed variables to latent common factors. In many situations, however, some outcome variables are in polytomous form while other outcomes are measured on a continuous scale. Maximum likelihood estimation for latent variable models with mixed polytomous and continuous outcomes is computationally intensive and may become difficult to implement in many applications. In this article, a computationally practical, yet efficient, Quasi- Maximum Likelihood approach for latent variable models with mixed continuous and polytomous variables is proposed. Asymptotic properties of the estimator are discussed. Simulation studies are conducted to examine the empirical behavior and to compare it with existing methods.