Background: Tyrosine kinase inhibitors (TKI) have revolutionized the treatment of gastrointestinal stromal tumours (GIST) although most patients develop resistance to first and second-line therapies. Regorafenib, an oral multi-targeted TKI, has demonstrated benefit in previously treated GIST patients. Methods: We assessed safety and preliminary activity of regorafenib in patients treated within the Managed Access Programme (MAP). All consecutive patients with advanced GIST who had progressed on or were intolerant of imatinib and sunitinib were recruited from the Royal Marsden Hospital and University College Hospital. We retrospectively reviewed the data for response, toxicity and treatment duration. Response was assessed by RECIST and Choi criteria. Toxicity was graded according to CTCAE v4.0 criteria. Results: 20 patients were included in the MAP in the UK between 3/2013 and 9/2013. Median age was 68 (range 45-87), 65% of patients were male. Performance Status was 0-1 for 18 patients (90%), 2 for 2 patients (10%). The median treatment duration was 29 weeks (range 1-50). 18 patients were assessable for response and all patients attained a best response of at least stable disease. At a median follow-up of 6.7 months, there were 2 partial responses (11%) by RECIST and 7 partial responses (39%) according to Choi’s criteria. One patient who had a PR on regorafenib had not benefited from previous sunitinib. 3 patients had disease progression and 3 patients discontinued treatment due to unacceptable toxicities; fistulation, myalgia and fatigue. 10 (50%) patients had grade 3 toxicities and 11 (55%) patients required a dose reduction. 5 patients started at reduced dose due to previous significant TKI toxicity, however 2 patients were able to be dose escalated. Median PFS and OS have not yet been reached but notably, prolonged stable disease was seen in 1 patient with exon 9 mutation and 1 patient with PDGFR D842V mutation. Conclusions: These data demonstrate encouraging activity of regorafenib in routine clinical practice. The documented adverse events are in line with previous trial data. Updated survival data and the role of RECIST/Choi response criteria in predicting survival will be presented.