The shelf-life of a drug product is the time that the average drug characte
ristic (e.g., potency) remains within an approved specification after manuf
acture. The United States Food and Drug Administration (FDA) requires indic
ation for every drug product of a shelf-life on the immediate container lab
el. Since the true shelf-life of a drug product is typically unknown, it ha
s to be estimated based on assay results of the drug characteristic from a
stability study usually conducted during the process of drug development. F
urthermore, the FDA requires that the estimated shelf-life be so constructe
d that it is statistically evident that the estimated shelf-life is less th
an the true shelf-life, i.e., the estimated shelf-life should be a conserva
tive (negatively biased) estimator. In this paper, we study and compare sev
eral shelf-life estimators, one of which is adopted by the FDA's 1987 guide
lines, in terms of their asymptotic biases and mean squared errors. Finite
sample performance of some shelf-life estimators is examined in a simulatio
n study.