ABSTRACT. An intensive study of a collection of arthropods from the Cretaceous of Lebanon, formerly referred to as stomatopod larvae, reveals that these forms belong to the problematical arthropod class Thylacocephala. The species Protozoea hilgendorfi, P. damesi, and Pseuderichthus cretaceus display defining thylacocephalan characters such as a carapace enclosing the entire body bearing a large anterior optic notch; three pairs of large, raptorial appendages; and a posterior battery of small swimming limbs associated with muscle segments. Unique characters of at least the genus Protozoea are the numerous 'pits' covering the entire carapace, elongated anterior rostral and posterior spines, and a dorsal hinge indicative of a truly bivalved carapace. We note several anatomical curiosities that provide some insight into the anatomy, ecology and phylogenetic relationships of these and other thylacocephalan species. Structures on the carapace surface of thylacocephalans may represent sensory organs comparable to the receptors seen on thecostracan crustaceans. The affinities of the three species described are still not entirely clear, but a phylogenetic analysis of the entire group may resolve this.T H Y L A C O C E P H A L A is a recently erected class of fossil arthropods, but species descriptions of what are now recognized as thylacocephalans go back over a century. Only recently have we come to realize the uniqueness of this group. Simultaneously, three independent groups of researchers described strange fossil arthropods whose unique characteristics forced them to erect a higher taxon to accommodate them. An Italian team worked on Ostenocaris cypriformis from the Jurassic of Osteno (Arduini et al. 1980;Pinna et al. 1982), which they assigned to a class of probable crustaceans they called 'Thylacocephala. ' In Paris, Secretan and Riou (1983) investigated specimens of the Jurassic Dollocaris ingens from France for which Secretan (1983) established the class 'Conchyliocarida. ' Finally, Briggs and Rolfe (1983), in their study of Devonian concavicarids, recognized a number of different arthropod genera within their new order 'Concavicarida.' Only after publication did it become clear that these arthropods all belonged to the same taxon. In 1985, Rolfe favoured the name Thylacocephala for the entire group (after Pinna et al. 1982), rendering 'Conchyliocarida' and 'Concavicarida' orders within this class, and for which Schram (1990) provided diagnoses. More and more species were added to the class by the three teams, together with workers like Polz (1989Polz ( , 1990Polz ( , 1994Polz ( , 1997 and Schram (1990). Most authors are uncertain about the exact affinities of this group, but it has been cautiously assumed that they are crustacean. However, we still lack irrefutable evidence for this.Although researchers agree that these forms all belong to the same group (even if the higher affinities remain uncertain), there are disagreements over the interpretation of several anatomical features (e.g. see Rolfe 1985). First...