OBJECTIVES: Brushing, the standard sampling method at endoscopic retrograde cholangiography (ERC), lacks sensitivity for cancer detection. We assessed a novel sampling method using a grasping basket. METHODS: Fifty-six patients with a suspected malignant biliary stricture were randomized to biliary sampling at ERC using a basket (basket group, N = 30) or a brush (brush group, N = 26), followed by the alternate device. When deemed necessary, strictures were dilated (using 6-mm balloons exclusively). The primary end point was sensitivity for cancer detection at cytopathological examination of the first sample collected in each patient; the cytopathologist was blinded to clinical details and sampling method. All analyses followed an intention-to-treat principle. RESULTS: All 56 patients had successful sampling with both techniques; 50 (89%) had a final diagnosis of malignant stricture. Sensitivity for cancer detection with the first sample collected in each patient was significantly higher in the basket compared to brush group (20/25 [80%]vs 12/25 [48%], respectively, P= 0.018, OR 4.33, 95% CI 1.24-15.21). Seventeen (34%) of the 50 sample pairs collected from malignant cases showed discordant cytopathological results: 15 patients had a positive basket and a negative brush result while two had the inverse association (P= 0.002, OR 7.5, 95% CI 1.65-47.44). Basketting more frequently yielded positive samples from malignant strictures in case of presampling balloon dilation (27/32 [84%]vs 10/18 [56%], respectively, P= 0.043, OR 4.32, 95% CI 1.14-16.37). Specificity was 100% (both methods). CONCLUSION: Biliary sampling at ERC using a dedicated basket provided a significantly higher sensitivity for cancer detection than brushing; presampling stricture dilation significantly increased sensitivity.