Endoscopic ultrasound (EUS)-guided fine-needle aspiration (EUS-FNA) is highly accurate in obtaining specific diagnoses for various diseases. The present study aimed to evaluate the diagnostic yields, accuracies and sampling adequacies, of slow-pull, 5 nil suction and 10 nil suction techniques in EUS-FNA of solid lesions. The present study was a retrospective comparative study, which was performed in tertiary academic centers, recognized for their expertise in PUS and PUS-guided FNA. The present study involved 149 patients who underwent EUS-FNA of solid masses. A wild of 34 (22.8%), 37 (24.8%) and 78 (52.4%) patients underwent EUS-FNA with slow-pull, 5 ml suction and 10 ml suction techniques, respectively. The EUS-FNA cyt6logy and histology results were compared with those from the gold standard of surgical histopathology [hematoxylin-cosin staining; immunohistochemical test of cluster of differentiation (CD) 79a, CD20 and flow cytometry test] or long-term clinical follow-up. The present retrospective comparative study demonstrated that the diagnostic yields and accuracies of EUS-FNA with slow-pull (86.1%) were significantly superior to those achieved with 5 ml suction (83.3%) or 10 nil suction (69.9%; P<0.0001; x2 test). Consistently, 86.5% (32/37) of the samples obtained from the 5 ml suction group were adequate for histological diagnosis. By contrast, 70.6 (24/34) and 85.9% (67/78) of samples from the slow-pull and 10 ml suction groups were adequate for histological diagnosis, respectively. The samples obtained using 10 ml suction contained more blood compared with those obtained via slow-pull and 5 ml suction (P=0.0056; y2 test). No complications were noted in any of the three groups. The samples that were obtained for histopathological diagnosis using 5 ml suction were superior to those obtained using slow-pull or 10 ml suction. Additional multi-central prospective studies in which EUS-FNA is performed with variable negative pressures arc required to improve the defining of the diagnostic roles of those techniques.