EUS-guided fine needle aspiration (EUS-FNA) is a highly accurate method of detecting malignant lymphadenopathy. The optimal methods for performing EUS-FNA to maximize sensitivity and to minimize the number of needle passes necessary are unknown. This is a report of the results of a prospective randomized controlled trial to determine the effect of suction, the site of FNA (edge or center of lymph node), and the method of preparation of cytologic specimens on accuracy, number of needle passes needed, and specimen quality. Consecutive patients with lymphadenopathy detected by EUS underwent FNA. Each lymph node was sampled with or without suction and from the edge or center in a 2 x 2 factorial design. The samples were expressed onto slides for cytology, and the residual material in the needle was analyzed by the cytospin-cellblock technique. Each aspirate was individually characterized for a diagnosis of malignancy, cellularity, and bloodiness. Forty-three patients with a total of 46 lymph nodes were evaluated. The final lymph node diagnosis was benign in 22 (48%), "suspicious for malignancy" in 6 (13%), and malignant in 18 (39%). The use of suction was associated with an increase in the cellularity of the specimen, but did not improve the likelihood of obtaining a correct diagnosis (OR 1.52: 95% CI [0.81, 2.85]). Samples obtained with suction were of worse quality because of excessive bloodiness (OR 4.7: 95% CI [1.99, 11.24]). Aspiration from the edge of the lymph node (compared with the center) did not increase the likelihood of a correct diagnosis (OR 1.16: 95% CI [0.42, 3.21]). For 78% of malignant lymph nodes, the correct diagnosis was obtained on the first needle pass and for 100% by the third pass. Cytospin-cellblock methods did not add any additional diagnostic information compared with direct smear cytology. The traditional method of applying suction during EUS-FNA does not improve diagnostic accuracy and worsens specimen bloodiness compared with FNA without suction. The site of FNA within the lymph node does not affect accuracy. When EUS-FNA is necessary, our recommendation is up to 3 FNAs without suction from the most convenient and safe location within abnormal-appearing lymph nodes.