This paper develops Bayesian approaches to deal with linear elliptical regression models that differ in the covariance structure. A pretest method based on posterior model probabilities is compared with a pooling approach, and the data density is defined as a mixture of elliptical densities with...