Opabinia regalis Walcott is an enigmatic fossil from the Middle Cambrian Burgess Shale of uncertain affinities. Recent suggestions place it in a clade with Anomalocaris Whiteaves from the Burgess Shale and Kerygmachela Budd from the Greenlandic Sirius Passet Fauna; these taxa have been interpreted as ‘lobopods’. Consideration of available Opabinia specimens demonstrates that reflective extensions from the axial region, previously thought to be either gut diverticula or musculature, can be accommodated in neither the trunk nor the lateral lobes that arise from it. They must therefore be external structures independent of the lateral lobes. On the basis of their sub‐triangular appearance, size and taphonomy, they are considered here to represent lobopod limbs. Some evidence for the existence of terminal claws is also presented. The question of whether Kerygmachela, Opabinia and Anomalocaris constitute a monophyletic or paraphyletic grouping is considered. While they share several characters, most of these are plesiomorphies. Further, Opabinia and Anomalocaris share several arthropod‐like characters not possessed by Kerygmachela. It is concluded that these three taxa probably form a paraphyletic grouping at the base of the arthropods. Retention of lobopod‐like characters within the group provides important documentation of the lobopod‐arthropod transition. A proper understanding of Opabinia and its close relatives, which may include the tardigrades, opens the way for a reconstruction of the arthropod stem‐group. This in turn allows the construction of a speculative but satisfying scenario for the evolution of major arthropod features, including the origin of the biramous limb, tergites and arthropod segmentation. ‘Arthropodization’ may thus be seen not to be a single event but a series of adaptive innovations.