White root disease (WRD) is one of the most serious infection problems in rubber plantation. At early infection stage, it is very difficult to diagnose the disease because infected trees do not exhibit any symptoms. Early detection of WRD is vital for farmers to initiate early treatment as well as to control the spread of the disease. Thus, this study investigated the feasibility of using spectroscopic method for early detection of WRD disease from rubber leaf samples. A total 50 rubber leaf samples representing five severity levels namely, healthy, light, moderate, severe and very severe infection were used in this study. Spectral data of the leaf samples was collected using visible shortwave near infrared (VSNIR) spectrometer. After the spectral measurement, chlorophyll content of the leaves was measured using SPAD meter. Partial least square (PLS) regression method was used to develop both calibration and prediction models for calibrating the spectral data with chlorophyll content. Artificial neural network (ANN) classifier was used to categorise the spectral data into the respective severity levels. This study found that the value of coefficient determination (R2) and root mean square error of calibration (RMSEC) were 0.99 and 0.56, respectively. For prediction model, the value of R2 and root mean square error of prediction (RMSEP) were 0.99 and 0.82, respectively. The ANN classifier yielded good classification accuracy of 90%. In conclusion, this study has provided a reliable basis to employ VSNIR for early detection of disease severity in rubber plantation.