The escalating interest in research on interventions that may affect a
ging processes has necessarily focussed attention on the outcome measu
res. The desirable characteristics of these ''biomarker variables'' ha
ve been widely discussed. This article offers some reflections on vali
dity, reliability, and generalizability of biomarkers. It is argued th
at our comprehension of aging will evolve iteratively from application
of a diversity of biomarker variables. Each of these will have streng
ths and shortcomings from methodological and measurement points of vie
w. The siren song that a ''gold standard'' index of aging can be found
should be ignored. Copyright (C) 1997 Elsevier Science Inc.