In many applications, data often exhibit specic behavior, namely unimodality, skewness and heavy tails. In this paper, we present a new, parsimonious family of distributions which models such features. The manner in which this is done, we show, provides explicit interpretation for all parameters in terms of location, shape, skewness and scale. This is important and relevant, in particular, to facilitate Bayesian inference. In addition to proving useful consistency properties, the theoretical ideas are applied to broad classes of models of interest to practitioners, using real and simulated data.