In a series of 135 patients with renal-cell carcinomas (followed up for a mean of 9.5 years), a variety of clinical and histological factors were analyzed in relation to morphometric measurements of the nuclear parameters in the primary tumours to establish their value as prognostic factors. Clinical, histological and morphometric factors were significantly interrelated in that the metastatic high-grade tumours had larger nuclei, larger variation in nuclear size and shape, and were also rapidly proliferating. In a univariate analysis, the most important clinical predictor of recurrence-free survival (RFS) was T category, followed by combined nuclear grade, N category, nuclear grade, tumour size, sex and M category. The most important quantitative predictor of RFS was the mean area of the 10 largest nuclei (NA10), the mean of the longest nuclear axis (Dmax), SD of nuclear area (SDNA), the volume-corrected mitotic index (M/V index), inflammatory-cell reaction, SD of nuclear perimetry (SDPE), and the mean of nuclear area (NA). M category, T category, combined nuclear grade, nuclear grade and N category were significantly related to patient survival. Of the quantitative variables, M/V index, Dmax and NA were significant predictors of survival in a univariate analysis. Females had longer RFS than men, and density of tumour-infiltrating lymphocytes (TIL) referred to an increased risk of recurrent tumour in both sexes. In a multivariate analysis, the RFS was independently predicted by the clinical stage, female sex and mitotic frequency/mm2, while nuclear parameters or nuclear grading had no independent prognostic value. The extent of the primary tumour was the single most important determinant of survival, followed by the proliferation rate of the tumour. In local T1-2NOMO tumours, mitotic frequency/mm2 was the only independent prognostic factor for RFS. The clinical stage, mitotic frequency/mm2, nuclear grade and density of TIL were independent predictors in Cox's analysis. In these local tumours, mitotic frequency/mm2 of neoplastic epithelium was the only independent prognostic factor. The results indicate that although an accurate prognostic evaluation of renal-cell carcinomas can be based on subjective nuclear grading and histoquantitative measurements of nuclear parameters, the simple assessment of mitotic frequency alone supplies most of the prognostic data, particularly in local tumours.