The paper discusses the types of aqueduct constructed by Roman engineers and outlines the evidence for their presence on military sites in Roman Britain. The evidence comprises not only discoveries of aqueduct channels, but also of water mains and fountains, both of which denote the presence of a piped water supply and are therefore strongly indicative of the existence of an aqueduct on the site. It is concluded that permanent fortresses were provided with aqueducts from the Claudian period, whereas they were only provided to supply permanent auxiliary forts from the Flavian period, probably as a consequence of the post-Neronian introduction of bath-houses.