BACKGROUND: One of the main challenges for clinical research in dengue is the low validity of clinical diagnosis. OBJECTIVE: To analyze clinical and laboratory data as predicitve factors of dengue diagnosis at Evandro Chagas Clinical Research Institute, Oswaldo Cruz Foundation, during the 2001-2002 dengue outbreak in Rio de Janeiro. METHODS: Cross sectional study comparing clinical laboratory data collected from the National Information System for Compulsory Notification Diseases (SINAN) in two serologically confirmed groups: dengue D (N = 453) and non-dengue ND (N = 80). RESULTS: Fever, exanthema, itching, mean platelet count < 150,000, WBC count < 4,000 and absence of vomiting and of abdominal pain help to distinguish D from ND groups. When considered individually, these signs and symptoms enhance diagnostic sensitivity and, when used in combination, improve specificity. CONCLUSION: A combination of symptoms not necessarily considered indicative of dengue diagnosis could improve surveillance and medical decision-making in simple clinical settings.