We report a four-year study of the dynamics of a population of Salvia lyrata L., growing in a greenhouse. The study exemplifies the population cage approach, which has been the basis of many long-term studies of dynamics of animal populations for over 50 years, but which has been overlooked by plant population biologists. The cohort of plants that founded the population had a very high probability of survival and, as adults, produced the greatest numbers of seeds and seedlings. Thereafter, the numbers of individuals in each stage class declined. This decline appeared to be a consequence of density-dependent processes. At the conclusion of the study, the population density was far lower than at the onset. The behaviour of the population differed from that of the population of S. lyrata studied in a nearby mown field in several respects, including timing of seedling emergence, average size of individuals and final density. Likely causes of these differences are discussed. We recommend the use of population cages in further studies of plant populations. Because it facilitates long-term observation, this approach can be of great use in elucidating the dynamic consequences of demography.