Background: There is a need for better control of endoscopic submucosal resection which is a minimally invasive method for removing small tumors, cancers or areas of dysplasia from gut mucosa or submucosa. It might also have an application in the treatment of Barratt s esophagus. Snare electrodiathermy injury patterns are often deeper than appreciated and saline injection though protective can be unpredictable. It is currently limited by difficulty in controlling the depth of the cut and by a consequent high risk of perforation. Our aim was to develop endoscopic devices which facilitate endoscopic submucosal resection to predetermined depths and decrease the risk of perforation. Methods : A device was constructed which could be mounted on the tip of an endoscope and arranged so that none of the functions were impaired and the field of endoscopic view was unimpaired in 3 out of 4 quadrants. The device featured a cavity into which tissue could be sucked. By adjusting the floating floor the depth of a tissue sucked into the cavity could be altered and different size resection capsules were constructed to fit the size of the tumor to be resected.Various methods of cutting of tissue were studied in combination with this instrument. Snare diathermy was designed to open in a rim of this cavity. (Shearing and blade cutting were also studied). The device was designed to be used in combination with endoscopic ultrasound and to incorporate a probe ultrasound transducer and allow real-time imaging of the tissue as the tumour is resected to a predetermined depth to ensure complete removal and avoid perforation. This device was tested on submucosal tumors created in postmortem stomach, oesophagus, colon. (Preliminary studies were performed in survival studies in pig.) Results : Artificial tumors were created by submucosal injection of saline, glues(including cyanoacrylate) and mucosa raised with elastic bands. The device successfully resected artificially produced submucosal and mucosal tumors at varying depths according to the adjustment of the floor of the suction cavity. When the device was used to remove small submucosal tumors created in post-mortem stomach perforation occurred 8/10 with the cavity depth set at 8mm, and 1/10 with the depth set at 5mm (p<0.001). Conclusion : This device allows precise control of depth of cutting during endoscopic submucosal resection of tumors and real time imaging of the layers of the gut during the cutting process. Dysplastic areas and Barrett's oesophagus could also be treated in this way.