OBJECTIVE: To investigate the relationship between the route to diagnosis, patient characteristics, treatment intent and 1 -year survival among patients with oesophagogastric (O-G) cancer. SETTING: Cohort study in 142 English NHS trusts and 30 cancer networks. PARTICIPANTS: Patients diagnosed with O-G cancer between October 2007 and June 2009. DESIGN: Prospective cohort study. Route to diagnosis defined as general practitioner (GP) referral-urgent (suspected cancer) or non-urgent, hospital consultant referral, or after an emergency admission. Logistic regression was used to estimate associations and adjust for differences in casemix. MAIN OUTCOME MEASURES: Proportion of patients diagnosed by route of diagnosis; proportion of patients selected for curative treatment; 1-year survival. RESULTS: Among 14 102 cancer patients, 66.3% were diagnosed after a GP referral, 16.4% after an emergency admission and 17.4% after a hospital consultant referral. Of the 9351 GP referrals, 68.8% were urgent. Compared to urgent GP referrals, a markedly lower proportion of patients diagnosed after emergency admission had a curative treatment plan (36% vs 16%; adjusted OR=0.62, 95% CI 0.52 to 0.74) and a lower proportion survived 1 year (43% vs 27%; OR 0.78; 95% CI 0.68 to 0.89). Urgency of GP referral did not affect treatment intent or survival. Routes to diagnosis varied across cancer networks, with the adjusted proportion of patients diagnosed after emergency admission ranging from 8.7 to 32.3%. CONCLUSIONS: Outcomes for cancer patients are worse if diagnosed after emergency admission. Primary care and hospital services should work together to reduce rates of diagnosis after emergency admission and the variation across cancer networks.