OBJECTIVE: To reassess PSA density (PSAD) in the detection of non-palpable, isoechoic prostate cancer and to analyze the effect of potentially inaccurate non-planimetric transrectal ultrasound volume estimates on the diagnostic performance of this diagnostic tool. METHODS: We prospectively evaluated 343 consecutive men with non-suspicious digital rectal examination and transrectal ultrasound findings and with serum PSA in the intermediate range (4.1-10 ng/mL). All men underwent systematic sextant biopsies of the peripheral zone. We performed a two-fold analysis of PSAD performance first using measured gland volume and then using modified gland volumes ranging from a 25% underestimation to a 25% overestimation, in 5% stepwise increments. RESULTS: With a 0.15 PSAD cut-off, we would have missed 14, 34 and 48% of cancers, if we had respectively used a volume underestimate of 25%, the measured volume and a volume overestimate of 25%. CONCLUSIONS: Potential gland volume under- or overestimation may substantially affect the diagnostic performance of PSAD. Although PSAD may represent a useful adjunct in the early detection of prostate cancer, it may compromise cancer detection in a substantial proportion of cases, if used as the only indicator for biopsy.