A size-classified transition matrix model of the demography of Arisaema triphyllum, a perennial herb of temperate deciduous forest, was constructed from 3 yr of field data from two dif~ere.nt sites. Ar!sa~n:za is quite unusual in that individuals change sex in a pattern determined by thetr stzes: small mdtvtduals do not flower, larger individuals reproduce as males, and the largest as females. As plants increase or decrease in size, they also may change sex.Though demographic parameters varied with the site and with the year, some consistent patterns were seen. Mortality rates ranged from 20 to 4()%,/yr for small, nonflowering plants, but decreased to ~uch lower levels as plants reached reproductive size. For these large plants, mortality rates did not dtffer between sexes or among size-classes. Because rates of sexual reproduction were severely pollinator limited, levels of seed production were low (averaging 7 seeds·female-'·yr-1 ) and were not correlated with plant size. All plants engaged in clonal reproduction, regardless of their size.. Transition matrices based on these data were compared statistically and were found to vary in ttme, as a result of year-to-year environmental variability. No population had achieved a stable size distribution. The classical method of calculating rate of increase, A., was therefore rejected, and an alternative method incorporating temporal stochasticity substituted. A sensitivity analysis was performed to evaluate the effect on population growth of changing various life history parameters. J:?ifferences in ra!es of individual growth and of clonal reproduction were found to be largely responsible for the relative success of the two populations, one of which was increasing, the other declining. Arisaema's status as a climax-forest herb suggests that temporally varying transition matrices might be common even in environments traditionally regarded as "stable." The prevalence of temporal variability has consequences both for the future direction of demographic theory and for the design and analysis of field studies.