We consider likelihood and Bayesian inferences for seemingly unrelated (linear) regressions for the joint multivariate t-error (e.g. Zellner, 1976) and the independent t-error (e.g. Maronna, 1976) models. For likelihood inference