To determine whether the first order kinetics model or the function proportional to the square root of disease duration better represents the progression of radiographic damage according to the disease duration than a linear model in rheumatoid patients. Two hundred twenty-eight patients assessed at a single time point were studied by using 2 nonlinear models of radiographic damage. In 44 patients a second radiographic analysis was taken within a 3-year period. In this data set, both models reveal the highest rate of progression in the first 3 years of the disease. The square root of the disease duration seems to better depict the whole disease course. Nonlinear models should be compulsory in the assessment of radiographic damage, especially when slow acting antirheumatic drug effects after 5 years of disease duration are studied.