Nine procellariiform species, covering a range of body mass exceeding 200 : 1, were studied during a visit to Bird Island, South Georgia, with the British Antarctic Survey, in the 1979-1980 field season. Speed measurements were made by ornithodolite of birds slope-soaring over land, birds flying over the sea but observed from land, and birds observed from a ship. In the second group, which showed the least anomalies, lift coefficients corresponding to mean airspeeds were about 1 for albatrosses, decreasing to about 0.3 for the smallest petrels. All species increased speed when flying against the wind. The small species proceeded by flap-gliding, while the large ones flapped infrequently, and only in light winds. The small species flew lower than the larger ones, but this may be related to the fact that most of the observations were of birds flying into wind. The albatrosses (Diomedea, Phoebetria) and giant petrels (Macronectes) were found to have a `shoulder lock', consisting of a tendon sheet associated with the pectoralis muscle, which restrained the wing from elevation above the horizontal. This arrangement was not seen in the smaller species, and was interpreted as an adaptation reducing the energy cost of gliding flight. The main soaring method in the large species appeared to be slope-soaring along waves. Windward `pullups' suggestive of the classical `dynamic soaring' technique were seen in large and medium-sized species. However, the calculated strength of the wind gradient would have been insufficient to maintain airspeed to the heights observed, and it was concluded that most of the energy for the pullups must come from kinetic energy, acquired by gliding along a wave in slope lift. Gliding downwind through the wind gradient should significantly increase the glide ratio, but this was not observed. Slope-soaring along moving waves in zero wind was recorded. The data were used to derive estimates of the average speeds that the different species should be able to maintain on foraging expeditions. Estimates of the rate of energy consumption were also made, taking into account the greater dependence on flapping in the smaller species, and on soaring in the larger ones. The distance travelled in consuming fuel equivalent to a given fraction of the body mass would seem to be very strongly dependent on mass. Comparison of the largest species (Diomedea exulans) with the smallest (Oceanites oceanicus) suggests that `range', defined in this way, varies as the 0.60 power of the mass, although the relation is more complex than a simple power function.