The Oriental fruit fly, Bactrocera dorsalis Hendel, is a serious pest insect for citrus fruits. The infected peel area can cause rot and fruit drop. However, there is no efficient automatic detection technology at this time that could detect the infected fruit. In this investigation, hyperspectral reflectance images were evaluated for detecting infected area on the citrus surface in the wavelength range between 500 and 950 nm. Optimum Index Factor (OIF) method was applied to identify the optimal band combination. Experiments result with a 97.5% recognition rate showed that hyperspectral imaging and proposed classification method were effective in differentiation of infected fruit and normal fruit. This study will lay a foundation for developing multispectral detection system used in on-line detection of infected fruit.