Population densities, reproduction, and survival of the lizard Uta stansburiana were measured at the Nevada Test Site in southern Nevada, USA, between 1964 and 1974. These data were used to develop a model of the population dynamics of this species. Results of irrigation experiments in 0.4—ha enclosures near Mercury, Nevada, were used to formulate multiple—regression equations predicting frequency and size of clutches laid by two age—classes of females in terms of winter rainfall, March air temperatures, and Uta population density. Densities of Uta in these enclosures were manipulated, and age—specific survival modeled in terms of spring densities of Uta. Experiments in which an important predator on Uta (the leopard lizard, Crotaphytus wislizeni) was removed from enclosures were used to estimate the influence of the predator on basic survival rates of hatchling and older Uta. The model was generally developed from data acquired in the small enclosures, but predictions were compared with actual observations of changes in Uta populations in Rock Valley (19 km west of Mercury, Nevada) between 1966 and 1972. Agreement between model predictions and actual numbers was fair. The model predicted a decrease in density from 1966 to 1967, but numbers of Uta actually increased conspicuously at this time. This was the only major discrepancy between predictions and observations. The observed mean spring density (d) between 1967 and 1972 was 41.4 Uta/ha (Sd = 20.8), while the model predicted a mean density of 37.8 Uta/ha (SD = 13.6). Observed and predicted mean proportions of yearlings in spring populations were identical (0.78). The basic version of the model estimated different survival rates for two age—groups of adult Uta. A simpler version of the model, using a common survival rate for both age—groups, gave predictions essentially identical with those of the basic model. Other tests of the basic model showed it to be most sensitive to changes in winter rainfall and predation pressure, much less so to air temperatures. Fifteen— and 30—yr synthetic sequences of predator densities were used to examine model stability over longer periods of time. When predator densities were drawn randomly from distributions with a mean of 2 individuals/ha, model populations exhibited lower mean numbers and amplitudes than actually observed during 9 yr in Rock Valley. The basic model included three density—dependent parameters: clutch frequency, clutch size, and adult survival. The model was modified so that (1) egg production was density independent, while adult survival was not; (2) adult survival was density independent, but egg production was not; and (3) there was no density dependence in the model. Thirty—year tests showed that cases 1 and 2 did not differ markedly from the basic model, although the removal of one density—dependent constraint resulted in slightly higher mean densities. In case 3, the model lacked stability and predicted numbers increased to unrealistic levels within 5 yr. We conclude that processes relating to egg production were modeled more effectively than those influencing survival, and that improvement of the model will depend on more detailed studies of the impact of predation on age—specific survival rates of Uta.