Studies attempting to document reproductive or other pre-lethal senescence effects in wild birds typically face an array of problems, including flaws in statistical analyses, non-adaptive philopatry to deteriorating environments, confounding effects arising from cohort heterogeneity and differential death rates of phenotypes and the frequent pairing of old birds to younger mates. Furthermore, recent studies suggest that birds could maintain a high level of physical fitness until old age, before being struck by a catastrophic illness leading quickly to their demise. The presence of terminally ill individuals in most datasets (and their greater incidence in older age categories) may therefore provide a false impression of progressive senescence in cross-sectional analyses. This study was designed explicitly to avoid all the known pitfalls linked to the demonstration of progressive senescence in wild populations, and involved one of the very longest-lived bird species. We show that, during incubation, old (aged 35 years and over) male grey-headed albatrosses Thalassarche chrysostoma make longer foraging trips, and have lower daily mass gains, than experienced mid-aged individuals (aged up to 28 years). This is, to our knowledge, the first report documenting reduced foraging performance with old age. Hatching and breeding success of pairs composed of two old individuals were reduced in comparison to mid-aged pairs. Overall results were very similar when analyses were repeated using only individuals known to have survived 1 or 2 years beyond field measurements (hence probably not suffering from the effects of an advanced terminal illness). We conclude that extremely long-lived individuals usually experience some degree of general physical deterioration, leading to reduced foraging and breeding performance, long before their final demise.