A theoretical investigation of the factors that affect the population dynamics of annual plants growing in deserts was conducted through the use of computer modeling techniques. A series of three models of the yearly life cycle of desert annuals was constructed and their behavior examined. The dissertation centers around the third and most complex model, a computer simulation model with distinguishable seed cohorts in a randomly varying rainfall environment. A typical simulation run was for 80 years and cost $1.00. The five plant functions were (l} seed losses (mainly predation) as a function of seed age, (2) seed dormancy as a function of seed ages (3) percent germination of the non-dormant seeds in response to germinating rainfall, (4) percent survival from the seedling stage to maturity as a function of total rainfall over the growing season and seedling density, and (5) seeds produced per p 1 ant as a function of total rainfall over the growing season and density of mature plants. The stochasitc rainfall generator used historical rainfall probabilities from US Weather Bureau stations at Las Vegas, Nevada and Tucson, Arizona. The literature on desert annuals was carefully searched to provide supporting data for the plant functions used in the simulation model. Most of the data is for winter annuals growing on the Nevada Test Site near Las Vegas. Single species data are rare, so the model functions reflected the average plant responses for winter annuals as a group. This base run set of functions reproduced the observed data quite well. Sensitivity analysis of the simulation model indicated that in order to persist in the Las Vegas area, the seeds of annuals should have at least a one-year period of dormancy and a minimum threshold of about 15 mm of germinating rainfall. The age distribution of the seed reserves in the soil and the percent germinable is strongly influenced by the recent rainfall history of the site and the seed loss rate. The optimum balance is when the losses of older seeds from the seed reserves due to germination is the same size as the sum of the non-productive losses (e.g., predation). Several experiments are suggested -- some to cover gaps in the published data and some that became evident through the sensitivity analysis of the model itself.