Résumé : We studied the influence of five cell nucleus populations taken as diploid standards with respect to the normalization of a human breast carcinoma. Four normal human tissues (lymphocytes, thyroid, liver, and bladder specimens) were taken as external standards, while the normal breast cells "contaminating" the tumor were taken as the internal diploid standard. Nuclear size and nuclear DNA assessments were performed by means of a cell image processor computing the parameters on Feulgen-stained nuclei from fresh imprint smears fixed in an ethanol-formalin-acetic acid mixture. Our results demonstrate that the choice of normal tissue as the diploid standard markedly influences the ploidy level of breast carcinoma. Normalization according to the lymphocytes led to our obtaining a major hyposextaploid G0-G1 DNA peak in the breast cancer. Using thyroid and liver cells as a standard, we obtained a major pentaploid and sextaploid G0-G1 peak, respectively. Using bladder cells or the normal contaminating breast cells within the tumor, we obtained a major tetraploid G0-G1 peak. Finally, the normalization of the normal bladder cells against the liver cells led to our obtaining a near triploid bladder specimen. The reverse feature was also observed, e.g., the obtaining of biologically nonsensical hypodiploid liver cells after normalization against the normal bladder cells. Such postnormalization variations in ploidy level depend upon the mean nuclear size and the mean nuclear DNA content of the normal tissue taken as diploid standard.(ABSTRACT TRUNCATED AT 250 WORDS)