Background: Zika is a new disease in the American continent and its surveillance is of utmost importance, especially because of its ability to cause neurological manifestations as Guillain-Barre syndrome and serious congenital malformations through vertical transmission. The detection of suspected cases by the surveillance system depends on the case definition adopted. As the laboratory diagnosis of Zika infection still relies on the use of expensive and complex molecular techniques with low sensitivity due to a narrow window of detection, most suspected cases are not confirmed by laboratory tests, mainly reserved for pregnant women and newborns. In this context, an accurate definition of a suspected Zika case is crucial in order for the surveillance system to gauge the magnitude of an epidemic. Methodology: We evaluated the accuracy of various Zika case definitions in a scenario where Dengue and Chikungunya viruses co-circulate. Signs and symptoms that best discriminated PCR confirmed Zika from other laboratory confirmed febrile or exanthematic diseases were identified to propose and test predictive models for Zika infection based on these clinical features. Results and discussion: Our derived score prediction model had the best performance because it demonstrated the highest sensitivity and specificity, 86_6% and 78_3%, respectively. This Zika case definition also had the highest values for auROC (0_903) and R2 (0_417), and the lowest Brier score 0_096. Conclusions: In areas where multiple arboviruses circulate, the presence of rash with pruritus or conjunctival hyperemia, without any other general clinical manifestations such as fever, petechia or anorexia is the best Zika case definition.