Background: We determined the incidence, probable risk factors, causes, and long-term survival of delirium in patients hospitalized in a medical specialty teaching hospital in Mexico City. Method: From June to December 1995, 667 elderly patients 60 years and older were hospitalized and assessed within 48 hours, excluding those with delirium at admission, those sedated, on respiratory support, or unable to speak. Results: Twelve percent of the population developed delirium, identified by means of the daily application of the Confusion Assessment Method; its appearance was attributed in 50% to two or more causes, in 10% to an insufficient control of pain, in 7.5% to a preceding surgical event, and in the rest to other causes. Each case was compared randomly with three nonpaired control patients of the same cohort who did not develop delirium. There was a significant increase in the number of cases of delirium in patients older than 75 years (p < .001), those with low schooling (p = .04), those with greater comorbidity (p < .001), those with a hematocrit lower than 30% (relative risk [RR] 2.1, confidence interval [CI] 1.2–4.1), and those with a glucose level greater than 140 mg/dl (RR 2.1, CI 1.2–3.6). Patients with delirium remained hospitalized longer than controls (p = .02). There was no significant difference in the intrahospital mortality of both groups, although during 5 years' follow-up, survivors demonstrated a significant increase in mortality (p = .03) in the group of individuals with delirium during the hospital stay when compared to controls. Conclusion: In this geratric polulation of Mexican patients, delirium incidence was similar to that previously reported in the worldwide literature. Its incidence is associated with longer hospital stay and greater mortality. Age, low level of schooling, greater comorbidity, high glucose levels, poor pain control, and hematocrit lower than 30% were independently associated with a greater incidence of delirium.