Pelomedusoides is the most diverse clade of side-necked turtles and there is an extensive fossil record (de Broin, 1988; Lapparent de Broin, 2000; Gaffney et al., 2006, 2011) that dates back at least to the Barremian (Lower Cretaceous) (Romano et al., 2014). Its large fossil record evidences a greater diversity in the past, particularly at the end of the Mesozoic, and exhibits a good sampling of species that are represented by skull material (Gaffney et al., 2006, 2011). As a consequence, the most complete and recent phylogenetic hypotheses for this clade (e.g. Romano et al., 2014; Cadena, 2015) are based on matrices comprising a great amount of cranial characters derived largely from Gaffney et al. (2006, 2011). In addition, it is well established that shell characters show a lot of phenotypic plasticity, even in the fossil species (Romano, 2008; Gaffney et al., 2006, 2011). In most cases it consequently is not justified to rely on “diagnostic features” of poorly informative shell-only material for describing a new species. Because of that, most authors remark new morphotypes in the literature when such aberrant specimens are recovered, but do not make any nomenclatural act by proposing a new yet poorly supported species (e.g. Romano et al., 2013; Ferreira & Langer, 2013; Menegazzo et al., 2015). Unfortunately, such a supposedly new bothremydid turtle (Pleurodira: Bothremydidae) from the Early Paleocene of Brazil was recently described based on poorly diagnostic remains (Carvalho et al., 2016; hereafter CGB, for the authors initials) and a correction of this unfounded nomenclatural act is required. In addition I present some comments on shell only material from Brazil in order to guide splitter-taxonomists to stop describing poorly preserved fossil specimens as new species.