This paper points out and discusses several practical and methodological pr
oblems that arise in attempts to detect and measure senescent declines in s
urvival or breeding performance of wild animals, with specific emphasis on
long-lived seabirds. Birds have no anatomical markers of age, so studies of
age-related biology require marking individuals at the time of hatching an
d following them throughout their lives. Seabirds live longer than the work
ing lifespan of biologists, and longer than the turnover times of study tec
hniques or theories of senescence. Seabirds are exposed to changing environ
mental and demographic conditions and cannot be assumed to be in demographi
c equilibrium. Sample sizes of the oldest age-classes are always small, req
uiring either marking very large numbers of birds at hatching or continuing
studies of old birds over many years. Incomplete sampling requires the use
of mark-recapture models that have only been developed in the last 20 year
s. Mortality selection resulting from demographic heterogeneity (selective
survival of high-quality individuals) can offset or confound the effects of
senescent changes within individuals. Many of these problems are amenable
to solution and will be probably solved within a few years. In the meantime
, this paper recommends that reviewers should be cautious about accepting p
ublished reports of senescent declines in natural populations. (C) 2001 Els
evier Science Inc. All rights reserved.