A phylogenetic analysis of the family Accipitridae was based on 188 osteological characters for 44 living genera, plus Haast's Eagle (Harpagornis moorei Haast, 1872), a large fossil species from the New Zealand Quaternary. Haast's Eagle is sister group to the Aquila eagles, which are themselves close to forest eagles of the genus Spizaetus. Most major groups recognised before were present, but some new groupings were revealed. Haast's Eagle is represented by copious material from over 40 sites, with more than 60 individuals. One 99% intact skeleton is known. A second nominal species (Hatpagornis assimilis Haast, 1874) is a junior synonym of H. moorei, and possibly represents the smaller male. The eagle's distribution did not match major environmental patterns, but appears to have been associated with a group of species of moas, Dinornithiformes. The eagle's distribution apparently changed with the climatic amelioration at the end of the Otiran glaciation, when it apparently retreated from northern and western areas as these became clothed in dense, wet forest. In the Holocene, it was most abundant in the east and south of the South Island, where there was a mosaic vegetation pattern of drier forest and shrublands. It was rare, or absent from inland and northern North Island districts during the Holocene. Sites in caves represent pit traps that caught eagles that entered to take live prey, probably large ground birds. Swamps may have trapped eagles that were attacking trapped moas and other birds, but the evidence is equivocal and specimens may represent natural attrition from a population. However, claw marks on 10% of moa pelves from birds in the 80-100 kg weight range in Canterbury Museum collections provide strong support for the predation hypothesis. The distribution of the eagle and its major potential prey species also support an hypothesis of an active predator rather than an obligate carrion eater. Ecomorphological analysis also supports the eagle's role as being a predator. Various multivariate statistical procedures consistently result in Haast's Eagle clustering with large forest eagles that use flapping flight, rather than with gliding eagles or vultures. This does not support the carrion feeder hypothesis. The eagle's wing proportions also suggest that it flapped rather than glided. There was some support for the two sexes having different flight patterns, and possibly different preferred prey. Haast's Eagle was the major predator in a mammal-free environment. Although phylogenetically an aquilin eagle, Haast's eagle had evolved into the largest, seemingly most powerful, forest and forest margin bird of prey known. The mosaic of features in this species illustrates the extreme plasticity, within narrow functional/historical limits, that characterises the Accipitridae.