We present the optical diagnosis of dengue virus infection in human blood serum using Raman spectroscopy. Raman spectra were acquired from 18 blood serum samples using a laser at 532 nm as the excitation source. A multivariate regression model based on partial least-squares regression is developed that uses Raman spectra to predict dengue infection with leave-one-sample-out cross validation. The prediction of dengue infection by our model yields correlation coefficient r2 values of 0.9998 between the predicted and reference clinical results. The model was tested for six unknown human blood sera and found to be 100% accurate in accordance with the clinical results.