Zika virus, an arbovirus responsible for major outbreaks, can cause serious health issues, such as neurological diseases. In the present study, different types of samples (serum, saliva, and urine), collected in 2015–2016 in New Caledonia and French Guiana from 53 patients presenting symptoms and clinical signs triggered by arbovirus infections, were analyzed using a recently developed, and in-house validated, 4-plex RT-qPCR TaqMan method for simultaneous detection and discrimination of the Zika and Chikungunya viruses. Subsequently, statistical analyses were performed in order to potentially establish recommendations regarding the choice of samples type to use for an efficient and early stage Zika infection diagnosis. On this basis, the use of only urine samples presented the highest probability to detect viral RNA from Zika virus. Moreover, such a probability was improved using both urine and saliva samples. Consequently, the added value of non-invasive samples, associated with a higher acceptance level for collection among patients, instead of serum samples, for the detection of Zika infections was illustrated.