A general theory for estimating the size of a closed population from m
ultiple-recapture data is presented. This theory is easily extended to
open population models for multiple-recapture data. Estimation is bas
ed on a log-linear model developed for modeling dependent capture-reca
pture data when capture probabilities vary temporally with behavioral
response and are heterogeneous among animals. Models for complete capt
ure history data are developed along with the first log-linear models
for removal data. The maximum likelihood estimator of the model parame
ters along with the estimated covariance matrix are presented.