The objective of our study was to estimate the hospital inpatient prevalence of diabetes mellitus in a Spanish tertiary care teaching hospital. We analyzed a cohort of 1036 patients consecutively admitted over a 7-day period to our hospital. We classified this total of hospitalized patients based on information obtained from individual analysis of medical history and values of plasma glucose after fasting, into groups with the following conditions: recognized diabetes, unrecognized diabetes, other hyperglycaemic situations, impaired fasting glucose (IFG) or non diabetes. One hundred and seventy-eight patients were estimated to have diabetes (total prevalence: 17.2%), including 158 patients with recognized diabetes and 20 patients with diabetes unrecognized before admission. Additionally, 25 patients were considered to have other hyperglycaemic situations and 20 patients were estimated to have IFG. The mean age of the diabetic patients was 65+/-13.7 years (50.5% men), and 94.4% had type 2 diabetes. Diabetes disproportionately affects the elderly inpatient, with a prevalence of 30.9% in people older than 64 years. Of the total number of patients with diabetes, only 144 (diabetes prevalence: 13.8%) were registered in hospital discharge records as having diabetes. We conclude that the extent of hospital diabetes prevalence considerably exceeds levels reported in the literature, suggesting that true diabetes prevalence in hospitals could be significantly under-reported, resulting in a serious underestimate of required expenditures.