Purpose: Many reports emphasize the importance of an accurate and reproducible method for measuring the ADC value of prostate tumors . Currently, ROI -based manual methods are the most commonly used quantitative approaches. The shapes of the ROI are drawn by many methods, such as circle, rectangle, contoured ( freehand) , and multi-circle ROIs. However, there is no consensus on what is the optimal method for measurement of prostate tumor ADC value. The purpose of this work was to investigate the influence and reproducibility of different-shaped ROIs on ADC measurements in prostate cancer, and to determine which method is the most consistent one. Material and Methods: Between April 2013 and October 2015, one hundred two patients ( age range, 46–92 years; mean, 63.5 years) with proven prostate cancer were evaluated in this retrospective study. Patients had undergone MR examinations including DWI using a 3.0T MR system. Two readers measured tumor ADC using four different ROI methods: circle ROI , rectangle ROI , contoured ROI , and multi-circle ROIs. Mean ADC and standard deviation were obtained. The intra- and interobserver reliability for measuring tumor ADC values by each ROI method were evaluated using intraclass correlation coefficient ( ICC) and Bland-Altman analyses. ADC values obtained with different ROI methods were assessed with a paired t test. Results: Mean ADC values were ( 0.747±0.129) ×10 mm /s for circle ROI , ( 0.743±0.129) ×10 mm /s for rectangle ROI , ( 0.758±0.127) ×10 mm /s for contoured ROI , and ( 0.745±0.135) ×10 mm /s for multi-circle ROIs. Statistical test showed that only contoured ROI featured asignificant high mean ADC value than any other shapes of ROI ( P<0.05) . Bland-Altman analysis displayed the ADC measurements obtained from the contoured ROI were more reproducible than those obtained from other ROI methods. The highest intra- and interobserver agreement was obtained for contoured ROI measurements, and the lowest was obtained for the rectangle ROI measurements. Conclusion: In conclusion, contoured ROI for outlined entire tumor size is more reliable for diffusion measurements and more appropriate clinically than use of arbitrarily determined smaller ROIs.