Perennial grass production on the Chihuahuan Desert Rangeland Research Center in south-central New Mexico was correlated with precipitation characteristics over a 34-year period. Total December through September precipitation was highly correlated (r =  0.77, n = 34) with perennial grass production. Practical generalized indices were developed that could be broadly applied by managers for predicting perennial grass production from precipitation characteristics. Perennial grass production and precipitation data on 3 separate pastures were collected over a 6-year period to evaluate the reliability of models to predict perennial grass production. Simple linear, 2-variable, quadratic, and polynomial regression models gave perennial grass production estimates that were well correlated with actual values (r =  0.85 to 0.91, n = 17) across the 3 pastures. The quadratic regression model (Y = 4.04 − 0.24X   0.012 X², X = December through September precipitation, Y = forage production, n = 34, r = 0.85) gave the most accurate predicted values. Our quadratic regression model should be of practical use to ranchers and range managers on Chihuahuan Desert upland rangelands receiving 200–300 mm annual precipitation, with loamy to sandy loam soils and in mid- to late-seral ecological condition. These conditions match those generally found on Chihuahuan Desert Uplands. We consider our quadratic regression model to be highly useful over large areas to ranchers in southern New Mexico, southeastern Arizona, southwestern Texas, and north-central Mexico.