The optimal method of magnetic resonance imaging (MRI)-ultrasound (US) fusion biopsy to adequately sample regions of interest (ROIs) remains unknown. We sought to determine the number and location of cores needed to adequately detect clinically significant prostate cancer (PCa). We identified patients undergoing MRI-US fusion prostate biopsy at our institution for known history or clinical suspicion of PCa. Multiparametric MRI studies were reviewed using Likert and Prostate Imaging Reporting and Data System (PI-RADS) v2 schema. Multiple targeted cores were taken from each ROI followed by 12-core systematic biopsy. In a distinct cohort of patients, lesions were targeted using a predetermined five-core template. We estimated cancers detected through sampling of five or fewer cores, assessed by core number and core location. We identified 744 patients with 581 lesions with PCa. Seventy-seven percent (279/361) of Gleason (G) ≥3+4 tumors and 72% (137/189) of G >3+4 tumors were detected on two-core sampling. Relative to all targeted cores, a two-core approach missed 16% of clinically significant cancers at first biopsy, 27% in prior negative, and 32% in active surveillance patients. Detection of G ≥3+4 cancers did not differ by core location. Sampling of two cores of ROIs misses nearly one-quarter of clinically significant PCa detected on additional sampling. PATIENT SUMMARY: We aimed to understand how the number of cores obtained from a suspicious area during prostate magnetic resonance imaging-ultrasound fusion biopsy affects cancer detection. We found that sampling of five cores missed substantially fewer cancers compared to two cores.