We initiated a prospective study of the dizzy patient to identify key factors on which a directed evaluation could be based. This study used a standardized history, physical examination, and basic laboratory evaluation totalling 66 items to assist collection of relevant clinical information on 125 patients. Diagnosis was based on the emergency physicians' diagnosis. This was modified when necessary based on one month of follow-up using diagnostic criteria adapted from previous studies. The most common disorder was some form of peripheral vestibular disorder, found in 54 patients (43%). These patients were typically vertiginous and were managed successfully as outpatients. Despite correlations with multiple factors, this diagnosis was best predicted by positive Nylen-Barany test with either vertigo, vomiting, or both with 94% specificity and 43% sensitivity. Potentially serious causes were identified, including medication-related, seizure, stroke, transient ischemic attacks, vertebral-basilar insufficiency, hypertension, pericarditis, arrhythmias, and all those requiring hospitalization. The best predictors, either older age, lack of vertigo, or neurologic deficit, could identify 86% of "serious" dizziness with 42% specificity. The following tests were of low yield and may be done in a directed manner based on a brief history: Valsalva, carotid stimulation, Romberg and Quix tests, mental status examination, complete blood count, serum electrolytes, and BUN. Our results do support routine testing of glucose in all patients and monitoring rhythm in patients age 45 and older. Such a directed approach could rapidly classify a significant number of dizzy patients and forego many time- and cost-intensive elements of provocative examination and laboratory testing.