Chronic lesions in renal biopsies are a well recognized prognostic factor for renal diseases, including lupus nephritis. The methods used for assessment of chronic lesions are, however, largely based on semiquantitative evaluation and may lead to poor reproducibility. Interobserver variation is particularly important in lupus nephritis, in which acute and chronic lesions may occur simultaneously. In this study we tested the reproducibility of chronic lesion assessment performed by three pathologists, two with specific training in renal pathology, using 20 renal biopsies and a standard semiquantitative method. In a second experiment, we evaluated the reproducibility of chronic lesion assessment in 33 biopsies of lupus nephritis by the two nephropathologists. The semiquantitative estimated values were compared with those from a previously proposed morphometric method for quantification of chronic lesions in renal biopsies. Although correlations were observed among the estimated values, there was a wide range of variation when semiquantitative methods were used. In particular, activity and chronicity indices of lupus nephritis were poorly reproducible. In contrast, use of a morphometric score, although not eliminating interobserver variability, led to better reproducibility of estimated values than that obtained with semiquantitative methods.