Transurethral resected tumor specimens of 61 patients with a primary and untreated bladder carcinoma were studied by selective nuclear morphometry, a method recently described by us. A significant enlargement of the mean nuclear area was found with the advance of tumor grade and stage (Wilcoxon, P less than .0001 and P less than .0001). The heterogeneity of the grade 2 patient group and the additional value of morphometry were demonstrated by observing the 5-year survival rates. Patients with grade 2 carcinoma could be separated into one subgroup with small nuclei (mean nuclear area less than or equal to 95 microns2) having a favorable outcome (5-year survival rate: 100%), and into another subgroup with large nuclei (mean nuclear area greater than 95 microns2) showing a worse prognosis (5-year survival rate: 63.2%) (Mantel-Cox, P = .01). The outcome of these subgroups was not significantly different from that of the grade 1 (5-year survival rate: 93.8%) and grade 3 (5-year survival rate: 50%) patients, respectively (Mantel-Cox, P = .45 and P = .57). The value of selective nuclear morphometry, in addition to tumor staging, was indicated by the association of nuclear enlargement (mean nuclear area greater than 95 microns2) with progressive recurrence (five of 15 patients; 33%) among the patients with conservatively treated superficial carcinoma (stages Ta and Tl). The findings demonstrate the supplementary value of selective nuclear morphometry to tumor grading and staging, especially in the heterogeneous group of grade 2 carcinomas and the group of superficial tumors (stages Ta and Tl).