AbstractCalculation of the correct nitrogen (N) fertilizer rate for a crop requires an estimate of the amount of N that mineralizes from soil organic matter during the growing season. A method proposed by Stanford and Smith for estimating N mineralization sometimes results in overpredictions which may be due to the use of disturbed soil samples. To study the effect of soil disturbance, we measured N mineralized in disturbed and undisturbed soil samples, and determined models that properly described the data for each type of sample. A double exponential model was required for disturbed samples, whereas a single exponential model was adequate for undisturbed samples. Although in general, models fitted to disturbed samples could not be used to predict N mineralized in undisturbed samples, they could be used to estimate the parameters of the single exponential model for undisturbed samples. Parameters estimated using this approach, together with daily values of soil temperature and water content, allowed reasonable predictions of the amounts of N mineralized in fallow plots on Haynie (coarse‐silty, mixed, calcareous, mesic Mollic Udifluvent) and Wymore (fine, montmorillonitic, mesic Aquic Argiudoll) soils. Parameters of the single exponential model for undisturbed samples could also be estimated from total N and clay contents of the soil. The results suggest that it may be possible to obtain accurate estimates of N mineralized in the field by estimating, instead of determining, the parameters of the single exponential model for undisturbed samples.