When the spontaneous mutation rate mu in mammalian cell cultures is estimated using Luria-Delbrück fluctuation analysis, many factors contribute to the unreliability of the estimate. Some of these have been documented by Featherstone et al. (1987) and by Kendal and Frost (1988). In particular, the plating efficiencies for mammalian cells are often much less than 100%, and this can be taken into account. A derivation of a generalized P0 estimator, mg, based on the classical P0 estimator of Luria and Delbrück (1943) is offered. In an experiment involving C cultures, of which z exhibit no mutant colonies, and in which the plating efficiency is p, (0 < p < 1), the estimated mean number of mutations per culture is given by [formula: see text] The classical P0 estimator is shown to be the limiting case of mg as plating efficiency tends to 100%.