Stat Methods Appl manuscript No. (will be inserted by the editor) Comments on: Model-based clustering and classification with non-normal mixture distributions L.A. Garc´ ıa-Escudero · A. Gordaliza · A. Mayo-Iscar Received: date / Accepted: date 1 Introduction First of all, we would like to congratulate S.X. Lee and G.J. McLachlan for this very stimulating work. The authors offer a motivating review of a quite new approaches to deal with clustering and classification problems by resorting to mixtures of non-normal distributions. Apart from presenting a systematic classification of multivariate skew-distributions (see also Lee and McLachlan 2013), they also review other alternative asymmetric mixture models that have been recently considered in the literature. Moreover, five nice real data appli- cations are also shown which make clear that the use of non-normal mixtures can be very useful in different real data problems where normality of mixture components is clearly not expected. In our opinion, these examples clearly show the increasing impact that non-normal mixture distributions will have in data modeling. In our comment, we will focus on two aspects. Firstly, we will discuss how trimming can be considered as an alternative way to handle outlying obser- vations. Secondly, we consider the issue of the addition of constraints on the scatter matrices in the formulation of these mixture problems to avoid degen- eracies and to avoid also the improper detection of non-interesting spurious solutions. L.A. Garc´ ıa-Escudero · A. Mayo-Iscar · A. Gordaliza IMUVA and Departamento de Estad´ ıstica e Investigaci´on Operativa. Facultad de Ciencias. Universidad de Valladolid. 47011, Valladolid. Spain. Tel.: +34-983-185878 Fax: +34-983-185861 E-mail: lagarcia@eio.uva.es