Harris et al. (2003) presented a lengthy discussion of problems of character construction in phylogenetic analysis, in part based on our (Heckert et al., 1996; Heckert and Lucas, 1999, 2000) cladistic analyses of aetosaur phylogeny. Their principal point with regard to character construction in aetosaurs is that certain characters of separate osteoderm regions (cervical paramedian, dorsal paramedian, lateral, and ventral osteoderms) should not be treated as independent characters because they covary. In terminology we prefer, they thus argue that these characters are redundant. Harris et al. preferred to call them intraorganismal homologues, making an unverifiable assumption about underlying genetic control of osteoderm morphology. When these redundancies are eliminated by combining various sets of the covarying osteoderm characters into one character, the tree obtained is still the same as that published by us (compare Harris et al., 2003: fig. 2a with Heckert and Lucas, 1999: fig. 9). Harris et al. then argued that their newly constructed “composite coding” tree (their fig. 2a), topologically similar trees (e.g., their fig. 2b), and our topologically similar tree (Heckert and Lucas, 1999; fig. 9) are weakly supported, based on various algorithms by which they estimated support. Not surprisingly, when they used fewer characters to describe some of the key morphological changes, the support declined for the nodes that these characters diagnose. We do not argue against the conclusion that aetosaur phylogeny is weakly supported, because we believe that aetosaurs are in general poorly understood. This poor understanding has little to do with analytical tools and much to do with the especially fragmentary nature of the aetosaurian fossil record. This problem was obvious to us even before we published an analysis with approximately 23% missing (unknown) data (Heckert and Lucas, 1999). We, our coworkers, and a growing number of fellow field researchers are attempting to rectify this problem by searching for new aetosaur fossils and redescribing the known specimens, many of which are fragmentary but diagnostic (e.g., Hunt and Lucas, 1990, 1991, 1992; Hunt et al., 1993; Heckert et al., 1996, 1999; Heckert and Lucas, 1998, 1999, 2000, 2002a, 2002b; Lucas and Heckert, 1996, 2001; Lucas, 1998; Lucas et al., 1998, 1999, 2002; Small, 1998, 2002; Zeigler et al., 2002). Not cited but also underway are specimen-based theses and dissertations in North and South America by Parker, Martz, and Desojo, preliminary results of which have been published in recent Society of Vertebrate Paleontology abstracts. It is abundantly obvious from these studies of new and existing collections that aetosaur skulls are, for example, much rarer than those of many contemporaneous taxa, including phytosaurs, rhynchosaurs, and dinosaurs. Aetosaurs apparently lived in environments that were not particularly favorable for preservation, resulting in an exceptionally fragmentary fossil record in desperate need of improvement. Furthermore, many of the studies cited above described fossils discovered, excavated, and collected by us, our coworkers, and volunteers as we try to improve this record. Our perspective, from the plaster-spattered highwall of one of our many excavations, is that we are far from the “upper bound” of better fossils, additional characters, and more character systems alluded to by Harris et al. (2003). Of the problems with aetosaur phylogeny that are related to analytical method, some of those Harris et al. (2003) pointed out are inherent to cladistic methodology, not just to our analysis. For example, there is actually a wide range of structures emanating from the dorsal (or lateral) surface of paramedian (or lateral) scutes, but in our analyses, we lumped them into apparently primitive keels (0) and presumably derived knobs (1). Consequently, our data matrix appears to have a bimodal distribution of many character states, when in fact the actual morphology is more variable. This problem is inherent to any attempt in cladistic analyses to quantify discretely the variation of essentially continuous characters. Now that we have examined many more aetosaurs and have collected new fossils of still others, we will address this variation in future analyses.