The assumption of equal catchability of marked and unmarked animals in mark—recapture sampling is examined. Violation of this assumption may come from two basic sources, heterogeneity or contagion, and both produce a bias in the estimate of population size. Two approaches to this problem are discussed. First, it is noted that bias can be avoided by statistically independent sampling procedures for marking and recapturing. Second, a new regression method is presented for multiple captures on a closed population, which combines mark—recapture with "simulated" removal and tests whether the assumption of equal catchability holds, regardless of the source of bias. Furthermore, if the ratio of catchabilities between unmarked and marked animals is constant over at least two successive samples, the method provides an estimate of population size even if the assumption of equal catchability does not hold. Two suggestions are made to supplement the regression method. First, noncapture sampling to obtain the ratio of unmarked to marked animals has several advantages over trapping. Second, a subclass of the marked animals may satisfy the constant ratio assumption even when all marked animals do not. A worked example is provided to illustrate the method.