Recent collections of sponges made in the Bahamas revealed the occurrence of two species of Agelas, which we could not assign to any of the currently recognised Agelas of the Western Atlantic. A comparison with literature records and extensive collections of Agelas incorporated in several major musea led us to conclude that several specimens belong to a species new to science, Agelas cerebrum. This species differs from the repent ramose form of Agelas conifera (Schmidt, 1870) in possessing a massively erect tube shape, thinner spicules and a higher number of whorls of spines. From the tubular form of Agelas conifera the new species differs in having much thicker tube walls with a system of convolutions and meandering grooves, and likewise differences in spicular dimensions. Agelas cerebrum differs from Agelas tubulata Lehnert & Van Soest, 1996 in having much thicker convoluted tubes and shorter spicules. A single thickly flabellate specimen we assign to Agelas dilatata Duchassaing & Michelotti, 1864, a species so far considered incertae sedis, because of compelling similarities with its original description and depiction. Our specimen differs from Agelas clathrodes in having a smooth plate form rather than an irregularly pitted wall-shape, and in having much more heavily cored primary fibres. It differs from Agelas inaequalis Pulitzer-Finali, 1986 because it is not cup-shaped and the spicules have more whorls of spines. It appears closest to Agelas flabelliformis (Carter, 1883) in shape and surface characteristics, but Carters species forms very thin blades of 5 mm, whereas Agelas dilatata is 25-30 mm thick. Agelas flabelliformis has distinctly shorter and thinner spicules.