Abstract pnutGRO is a peanut crop growth simulation model which calculates crop carbon, nitrogen and water balances at the process level. An on-farm evaluation of pnutGRO v1.02 was conducted at 15 field sites in two Florida counties during the 1990 and 1991 cropping seasons. Independent crop and soil data sets were collected to evaluate pnutGRO simulations. The accuracy of the pnutGRO simulations was affected by year and location. Sites where peanut (Arachis hypogaea L.) was grown in rotations following bahiagrass (Paspalum notatum Fleuge) (Levy County, 1990) had low disease pressure, high pod yields (5260 kg/ha), and the best model fit (pnutGRO simulations were within 9% of observed yield data). Sites where peanut followed other row crops often showed high infestation levels of root-knot nematode [Meloidogyne arenaria (Neal) Chitwood] or stem rot (Sclerotium rolfsii Sacc.) (Jackson County, 1991). This resulted in reduced pod yields (3260 kg/ha) and poorer model fit (pnutGRO simulations were 44% above observed yields). pnutGRO correctly predicted relative yield decreases due to drought. Overall, pnutGRO v1.02 appears to be most useful as a predictor of optimal peanut yield for specific cultivars under given soil and weather conditions. Inclusion of pest and disease damage functions would improve model accuracy for farms where biotic stress reduces potential peanut yield.