A study has been made of the correlation between objectively measured nuclear DNA content (by image analysis cytometry) and visual pathologic/cytologic diagnoses in hyperplasia/inflammation and well, moderately and poorly differentiated carcinomas of the prostate. The DNA measurement data were described by four statistical description methods--namely mean deviation from 2C, percentage greater than 4.25 C, DNA index and entropy. In 36% (5/14) of well differentiated and 13% (1/8) of moderately differentiated cancer it was not possible by objective measurement methods to demonstrate any difference between hyperplasia/inflammation and cancer. On the other hand one or more of the statistical descriptors was significantly different from hyperplasia/inflammation in all of those cancers visually diagnosed as poorly differentiated. The results underline the need for improvement of diagnostic criteria of prostate cancer--in particular well differentiated--in order to correctly understand the epidemiology (true incidence), survival and effects of treatment on the disease.