The main objective of this cohort study was to analyze the influence of social factors on the interval from the onset of symptoms to diagnosis in the 217 patients with digestive tract cancer. From the clinical charts, the following information was extracted: date of the beginning of illness, dates of medical attendance, type and date of diagnostic tests, characteristics of the disease (symptoms, stage), and date of treatment. From personal interviews we gathered the data not present in the clinical chart: sociodemographic data, information about other underlying diseases, and variables related to lifestyle. The time periods considered were from onset of symptoms to initial consultation (S-C); from initial consultation to hospital admission (C-H); and from hospital admission to diagnosis (H-D). The information on hospital admission, diagnosis, and treatment was prospectively gathered, whereas data before hospitalization were assessed retrospectively. For statistical analysis, we used Kruskal-Wallis test and Cox regression. Pain was the only predictor associated with a shorter S-C period. Hemorrhage, number of symptoms, and hospital consultation significantly shortened the interval C-H. Both hemorrhage and pain, and an initial visit at private primary care were the independent predictors that decreased the period H-D. In conclusion, sociodemographic factors did not influence healthcare attendance.