1 Lefkovitch transition matrices were used to determine vital demographic rates of a natural population of Taraxacum officinale in Morgantown, WV, USA. Separate size transition matrices were calculated for each of four seasons, October-January, January--April, April--July, and July--October, to test if demographic rates vary as a function of season and if size-specific rates vary differentially among seasons. Season-dependent demography was also compared for four phenotype classes segregated by cluster analysis of leaf morphology. 2 The finite rate of increase for the entire population was largest in autumn (October-January) and declined throughout the rest of the year. Overall, there was a small reduction in the population size. Size-specific probabilities of survival, growth and fertility varied dramatically among seasons. Sensitivity analyses showed that small individuals were particularly important to population growth from autumn to spring. Larger individuals were more important during summer. 3 Highly season-dependent demographic rates have large implications for population distribution and persistence since increased vulnerability to perturbation during particular seasons may constrain population growth and stability. Although T. officinale is a long-lived perennial, annual censuses may mask the importance of certain individuals or life history traits for maintenance of genetic variability and population viability. 4 Seasonal and annual finite rates of increase also varied as a function of phenotype class. Of two phenotype classes which had identical annual growth rates, one grew better in cool seasons while the second performed better in warm seasons. Direct competition for resources should be reduced by such inverse patterns of demography across seasons. 5 If phenotype classes are to some degree genetically determined, the differential responses observed here suggest that temporal variation in the environment could explain the maintenance of genetic diversity within populations.