Fitness is expected to decrease with inbreeding in proportion to the amount
of deleterious genetic variation present in a population. The effect of in
breeding on survivorship is usually modeled as a negative exponential relat
ionship, and this model has been widely used to estimate the amount of dele
terious genetic variation in populations. Linear regression has traditional
ly been used to estimate the parameters of the: model, including the number
of lethal equivalents. This article describes an alternative method for es
timating parameters and their confidence limits: the maximum likelihood app
roach. The accuracy of regression and maximum likelihood estimates of the n
umber of lethal equivalents is compared through simulation. The maximum lik
elihood approach is found to be both median unbiased and capable of estimat
ing confidence limits with nearly the stated degree of accuracy, while the
linear regression approach is found to be median biased. The significance o
f this on previous estimates of inbreeding depression is discussed. (C) 199
8 Wiley-Liss, Inc.