The objective of this research was to evaluate those morphological and anatomical characters and their states that have been conventionally described in systematic studies of octopuses, focussing on those of the family Octopodidae. Additional characters that may assist in determining both phylogenetic relationships between taxa, and differentiation of taxa in this family, were also examined. To undertake this research, representatives of a variety of octopus genera were examined, primarily incirrate Antarctic eledonids (characterised by having a single row of suckers along each arm), such as species of Adelieledone, Bentheledone, Graneledone, Megaleledone, Pareledone and Thaumeledone, augmented with species with two rows of suckers, an Antarctic species of Muusoctopus, and non-Antarctic species of Enteroctopus, Octopus and Pinnoctopus; one cirrate, Cirroctopus, was also examined. The taxonomic and systematic status of a number of these genera, and even those species assigned to them, has not been completely resolved, or they have been the subject of recent debate. For each species, conventional characters such as the alimentary canal, female and male reproductive systems, and upper and lower beaks, are described and figured, augmented with less conventional descriptions of the ink sac, such as the extent to which it is embedded within the digestive gland, its relative size, and, when possible, the major branches of the arterial system leading from the heart to the reproductive, renal and alimentary systems, and ink sac. To assess the usefulness of morphological characters frequently used in octopodid classifications, multivariate statistical analyses were performed on data sets of measurements of external and internal anatomy. Four data sets were analysed separately, then collectively using Principal Component Analysis (PCA) and Multidimensional Scaling (MDS): 1) external characteristics, including those cited as basic to octopus descriptions (Roper & Voss 1983); 2) internal characters of the alimentary canal and mantle cavity configuration; and 3) lower, and 4) upper beak morphologies. As indicated by PCA and MDS, the use of external characteristics only groups taxa into intuitively unnatural groups of grounds on statistical similarity between measures, states and categorical characters. In contrast, PCA and MDS results for internal and external characters group taxa in an intuitively more natural manner. Descriptions of octopus taxa as a consequence should describe both. Beak morphology is of more limited value for discriminating closely-related species; however, in some cases it does add information of taxonomic importance, so descriptions of these structures are recommended. A subset of those characters that best differentiate genera and species are identified. At the generic level these comprise arm sucker counts (ASC), arm length index (ALI) and head width and length indices (HWI, HdLI) for externals; oesophagus length index (OesTLI), anterior and posterior salivary gland length indices (ASGLI, PSGLI), intestine length index (IntLI) and anterior oesophagus index (AoesLI) for internals; beak length and crest height indices (BLI, CHI) for lower beak, and rostrum width and wing length indices (RWI, WLI) for upper beak. At the specific level these comprise free funnel and funnel length indices (FFI, FLI) for externals; anterior oesophagus length index (AoesLI), posterior salivary gland length index (PSGLI) and gill count (GilC) for internals; crest height index (CHI) for lower beak, rostrum beak length index (BLI), rostrum edge and width indices (REI, RWI) for both beaks, and Wing length index (WiLI) for the upper beak. Phylogenetic relationships between the various taxa analysed herein are not assessed because the statistical techniques employed, PCA and MDS, are not appropriate tools to use for phylogenetic reconstruction. More studies about which characters are useful for phylogeny need to be undertaken to improve classification. Characters that have been used extensively to differentiate subfamilies of octopodids (ink sac and number of sucker rows) are of limited phylogenetic value, and as such their use to define subfamilies should be avoided, treated with extreme caution, or the concept of the subfamily should be avoided altogether until relationships are independently corroborated by independent molecular data.