Background: 90% of polyps removed at colonoscopy are <10mm in size and only 50% of these are neoplastic. The number of neoplastic polyps is the key determinant of future surveillance intervals and small adenomas, especially those less than 5 mm in size, rarely contain advanced pathology. Accurate characterisation of small polyps in vivo could reduce the number of unnecessary polypectomies, reduce the need for polyp retrieval and greatly reduce histopathology costs. Accurate characterisation of small polyps in vivo is feasible, with a number of studies reporting accuracy around 90% when using magnification, chromoendoscopy or narrow band imaging (NBI). However the ability to make clinical decisions based on optical biopsy has not been investigated. Aims and Methods: This prospective study examined whether colonoscopists could accurately assess small polyps in vivo and determine appropriate surveillance intervals based on real time optical biopsy. Consecutive patients attending for colonoscopy (FOBT positive or adenoma surveillance) were included. Four endoscopists with varying experience of optical biopsy performed all the colonoscopies. Once a polyp < 10 mm was encountered the endoscopist predicted polyp histology using one or a combination of optical modalities (high definition white light alone, NBI or indigocarmine chromoendoscopy). At the end of the procedure, future surveillance intervals based on optical biopsies were determined according to BSG guidelines. Optical diagnosis was compared to histopathology findings. The primary outcome measure was the accuracy of optical polyp characterisation and clinical decision making (ie to resect or leave polyps in situ). A secondary outcome was any change in surveillance interval following histopathological assessment. Results: 168 (out of 296 planned) polyps smaller than 10 mm (mean size 3.5 mm, range 1-9 mm SD 2.2) were found in 57 patients. Histology was available for 141 lesions - 85 adenomas, 38 hyperplastic polyps, 1 inflammatory, 3 lymphoid follicles and 14 normal. 92% (78/85) of adenomas and 95% (36/38) of hyperplastic polyps were correctly diagnosed by optical biopsy. In 82/86 (95%) cases the decision to resect and discard the polyp could be considered correct. Two adenomas contained high-grade dysplasia - in both of these cases, the colonoscopist predicted the diagnosis of adenoma and elected to resect and discard but the subsequent colonoscopy surveillance interval was unaffected. There were no cancers smaller than 10mm. Following BSG Guidelines for polyp surveillance, 0/57 patients had their predicted surveillance interval changed after formal histopathology.