The objective of this study was to compare the pre-hospital health care process, clinical characteristics at admission and survival of patients with a digestive tract cancer first admitted to hospital either electively or via the emergency department. The study involved cross-sectional analysis of information elicited through personal interview and prospective follow-up. The setting was a 450-bed public teaching hospital primarily serving a low-income area of Barcelona, Catalonia, Spain. Two hundred and forty-eight symptomatic patients were studied, who had cancer of the oesophagus (n = 31), stomach (n = 70), colon (n = 82) and rectum (n = 65). The main outcome measures were stage, type and intention of treatment and time elapsed from admission to surgery; the relative risk of death was calculated using Cox's regression. There were 161 (65%) patients admitted via the emergency department and 87 (35%) electively. The type of physician seen at the first pre-hospital visit had more often been a general practitioner in the emergency than in the elective group (89% vs 75%, P < 0.01). Emergency patients had seen a lower number of physicians from symptom onset until admission, but two-thirds had made repeated visits to a primary care physician. Emergency patients were less likely to have a localized tumour and a diagnosis of cancer at admission, and surgery as the initial treatment. Median survival was 30 months for elective patients and 8 months for emergency patients (P < 0.001), and the relative risk of death (RR) was 1.83 (95% confidence interval, CI, 1.32-2.54). After adjustment for strong prognostic factors, emergency patients continued to experience a significant excess risk (RR = 1.58; CI 1.10-2.27). In conclusion, in digestive tract cancers, admission to hospital via the emergency department is a clinically important marker of a poorer prognosis. Emergency departments can only partly counterbalance deficiencies in the effectiveness of and integration among the different levels of the health system.