Population dynamics of Black-browed and Grey-headed Albatrosses were studied at Bird Island, South Georgia for 17 consecutive years (1975-1991). Over this period almost all the Grey-headed Albatross colonies decreased, at an average rate of 1.8% per annum. Although the total Black-browed Albatross population increased (at 0.8% p.a.), 14 of the 23 colonies (including both study colonies) decreased. Black-browed Albatrosses follow an annual breeding cycle, with over 80% of birds successful in rearing a chick and 75% of those failing to do so returning to breed the next year, 5- 10% of both categories delaying one further year (even when still paired). Grey-headed Albatrosses are essentially biennial, <1% of successful birds breeding the next year, 68% returning 2 years later, 11 % the next and 5% not until the fourth year. In contrast, over 50% of birds failing to rear a chick breed the next year, 23% delaying for one further year. This largely reflects that birds failing after March do not return the next year, whereas 80% of birds failing during incubation do breed the following year. Grey-headed Albatrosses show higher (39%) and more consistent breeding success than Black-browed Albatrosses (29%, including 4 years of almost complete breeding failure); hatching success is similar in both species and fledging success is the main source of variation in Black-browed Albatross productivity. We link this to the dependence of this species on Antarctic krill, a variable resource largely absent in 3 of the 4 years of widespread failure. Modal age of first breeding is 10 and 12 years for Black-browed and Grey-headed Albatrosses respectively. Overall, survival rate of juveniles to recruitment as breeding birds averaged 28% and 14% for 1960's and 1970's cohorts, respectively, of Black-browed Albatrosses; for Grey-headed Albatrosses similar values were 38% and 6%. Studies of immigration and emigration between colonies show that no breeding birds move, that philopatry of fledglings is generally high but that recruitment rates need to be increased by 1-3% to allow for juveniles returning to non-natal colonies. Annual survival of adults averages 93%, s.e. = 8 and 95% (s.e. = 0.8) for Black-browed and Grey-headed Albatrosses respectively: for sexed birds. rates are similar for Grey-headed Albatrosses but 2% higher for female Black-browed Albatrosses. With our data lifetime productivity would be 30% higher for Black-browed Albatrosses but this could be offset by a change in adult survival rate of less than 1%. The proximate reason for the population decline in studied colonies is high juvenile mortality, which has increased since the 1960's. It is likely that this reflects incidental mortality associated with fisheries (particularly entanglement in long-lines for tuna and collision with net-monitor cables of trawlers) but data are few and mainly circumstantial. Many differences between Grey-headed and Black-browed Albatrosses (e.g. the former showing later sexual maturity, less frequent breeding. higher and more consistent survival) probably relate to basic differences between annual and biennial breeding. Some differences (e.g. breeding success) probably relate to differences in diet; others (e.g. in juvenile survival) may reflect different at-sea distributions. Comparisons with other species reveal species and possibly site-specific differences rather than consistent patterns for annual and biennial breeders.