Basal stem rot (BSR) caused by Ganoderma boninense is known as the most destructive disease of oil palm plantations in Southeast Asia. Ganoderma could potentially reduce the market share of palm oil for Malaysia. Currently Malaysia produces about 50% of the world’s supply of palm oil. Early, accurate, and non-destructive diagnosis of Ganoderma fungal infection is critical for management of this disease. Early disease management of Ganoderma could also prevent great losses in production and potentially reduce the use of chemicals. In this study, we propose to apply a mid-infrared spectroscopy technique for detection of infected oil palm trees at three stages of infection. Leaf samples of healthy, mild, moderate and sever-infected trees were measured using Fourier transform infrared (FTIR) spectrometers system to obtain absorbance data from the range of 2.5-25 µm. Single bounce ATR accessory with dilution with KBr were used in this study. Savitsky-Golay method was used for smoothing. The selected principal component (PC) scores were used as input features in linear discriminant analysis (LDA) as a pattern recognition algorithm. The results indicated that LDA-based algorithm can distinguish between healthy and infected leaves at three stages of infection with high classification accuracies (˃90%). Thus, this study demonstrated that the proposed technique has the potential for early detection of the Ganoderma disease.