Early detection of fungal infections in citrus fruit still remains one of the major problems in postharvest technology. The potential of laser-light backscattering imaging was evaluated for detecting decay in citrus fruit after infection with the pathogen Penicillium digitatum, before the appearance of fruiting structures (green mould). Backscattering images of oranges cv. Navelate with and without decay were obtained using diode lasers emitting at five different wavelengths in the visible and near infrared range for addressing the absorption of fruit carotenoids, chlorophylls and water/carbohydrates. The apparent region of backscattered photons captured by a camera had radial symmetry with respect to the incident point of the light, being reduced to a one-dimensional profile after radial averaging. The Gaussian–Lorentzian cross product (GL) distribution function with five independent parameters described radial profiles accurately with average R2 values higher or equal to 0.998, pointing to differences in the parameters at the five wavelengths between sound and decaying oranges. The GL parameters at each wavelength were used as input vectors for classifying samples into sound and decaying oranges using a supervised classifier based on linear discriminant analysis. Ranking and combination of the laser wavelengths in terms of their contribution to the detection of decay resulted in the minimum detection average success rate of 80.4%, which was obtained using laser light at 532 nm that addresses differences in scattering properties of the infected tissue and carotenoid contents. However, the best results were achieved using the five laser wavelengths, increasing the classifier average success rate up to 96.1%. The results highlight the potential of laser-light backscattering imaging for advanced citrus grading.