We assessed the sensitivity and the positive predictive value (PPV) of the ICD-9 codes in identifying ischemic strokes. The study involved the cross-sectional comparison between patients with an ischemic stroke diagnosis made by neurologists and patients with the 434 or 436 discharge codes. Sensitivity of the codes (all diagnostic levels and first level respectively) was 82% and 76%; PPV: 71% and 76%. The annual crude incidence of ischemic stroke was 2.62 per 1000 based on verified strokes and 3.03 per 1000 based on 434 or 436 coded medical records (at all diagnostic levels). Thirty-day case fatality ratio was 22.3% in verified strokes and 36.8% among patients diagnosed with codes 434 or 436 but without stroke (all levels). Our results disclosed inaccuracy in use of the ICD-9 codes in the diagnosis of ischemic stroke in the general hospital of Lugo di Romagna, Ravenna Province, Italy. The misdiagnosis of patients could be influenced by the degree of severity of clinical features. Epidemiological data and cost-analysis forecasts based only on the ICD-9 system must be considered with caution.