Background: Many pts with aGOAC are elderly and/or frail. We previously compared epirubin/ oxaliplatin/ capecitabine (EOCap) vs OCap vs Cap in a pick-the-winner study and found OCap best. GO2 was designed to find the optimum dose of OCap and to explore the use of an objective baseline geriatric assessment to individualize doses for maximum Overall Treatment Utility (OTU), a composite of clinical benefit, tolerability, QL and patient value. Methods: Pts with aGOAC were eligible if unsuitable for full-dose EOCap due to age or frailty, but fit for OCap; GFR ≥ 30, bili <2x ULN. Baseline assessment included global QL; symptoms; functional scales; comorbidity; frailty. Randomization was 1:1:1 to dose Level A (Ox 130 mg/m2d1, Cap 625 mg/m2bd d1-21, q21d), B (80% Level A doses) or C (60% Level A doses). Pts with GFR 30-50 ml/min or bili 1.5-2.0 xULN received 75% of the allocated dose of Cap. At 9 wks, pts were scored for OTU. Continuation thereafter was based on clinical judgement. Non-inferiority (vs A) was assessed using PFS censored at 12 months, with boundary HR 1.34 (based on discussion with pts and clinicians), needing 284 PFS events per 2-way comparison. Baseline fitness was assessed as predictive of OTU, overall and by interaction with dose level. Results: 514 pts were randomised, 2014-17, at 61 UK centres. Clinical trial information: 44687907. Non-inferiority of PFS is confirmed for Level B vs A (HR 1.09, CI 0.89-1.32) and for Level C vs A (HR 1.10, CI 0.90-1.33). Level C pts had less toxicity and better OTU outcomes than A or B. When analysed by baseline age, frailty and PS, Level C produced the best OTU even in younger, less frail and better PS patients; no group was identified who benefit more from the higher dose levels. Conclusions: This is the largest RCT to date specifically investigating frail and/or elderly aGOAC pts, and should guide future treatment. The lowest dose tested was non-inferior in terms of PFS and produced less toxicity and better overall treatment utility.