Ploidy analysis is an aid in the diagnosis and evaluation of prognosis of tumors. Image analysis is a relatively simple technique to assess ploidy that can be carried out with accessible equipment. However, it requires the use of accurate technical procedures to avoid methodological errors that may bias the measurements. We previously developed two procedures that are simple to apply in routine work and serve to correct the errors derived from the measurement of large nuclei that are not fully contained in the thickness of the section and those derived from non-specific background readings. In the present study we applied these corrections to the retrospective ploidy analysis of a series of 67 patients with oral carcinoma with a follow-up time of 18 months. Thirty-four patients were alive at the end of the study, 33 were deceased. The ploidy values and the malignancy indices corresponding to the deceased and live patients with TNM stage III and IV carcinomas at the time of biopsy were significantly different. There were no significant differences in ploidy values between live and deceased patients with TNM stage I and II at the time of biopsy. The corrections improved the sensitivity of the method and thus the statistical significance of the data. These data suggest that the method proposed may be of use to estimate lesion evolution, in particular in patients with advanced oral squamous cell carcinomas.