Estimates of population density are often used to assess recovery actions for listed species such as the desert tortoise, (Gopherus agassizii) for which distance sampling is used to estimate population density. However, distance sampling requires precise estimates of the availability of individuals to be sampled (G0) and the probability of detecting animals (Pa). G0 for desert tortoise has been estimated by intensively observing small populations of radio-telemetered animals throughout the sampling season. The sizes of these intensively observed populations previously have been shown to be too small to provide precise estimates of sampling availability. To reduce the variance in the estimate of G0, we used small dataloggers mounted on animals to measure the microclimate experienced by individuals as a way to model the activity of individuals and their availability to be sampled throughout the season. The resulting estimates of G0 were used, in conjunction with reported detection probabilities, to calculate population density. Additionally, we simulated variance in each of the parameters as being the same as the measured variances from the field for G0 and Pa to assess the influence of these parameters on the precision of population density estimates. Coefficients of variation for estimates of population density using realistic ranges of G0 and Pa parameters indicate that distance sampling has insufficient precision to detect trends in population size over time of less than 3% change per year as estimated from data over 25 years. If management actions are to be assessed using estimates of population density, alternatives will be needed.