This paper develops a family of simulators of the multivariate t proba
bility, based on a suitable choice of an importance sampling distribut
ion. These simulators are useful for the evaluation of posterior densi
ties and for hypothesis testing in Bayesian analysis.