Cydia pomonella (L.) and Cydia molesta (Busck) (Lepidoptera: Tortricidae) are two important lepidopteran pests that may co-occur in apple orchards and are difficult to differentiate in the larval stage. We investigate the possibility of using near-infrared spectroscopy (NIRS) coupled with partial least squares analysis to distinguish the larvae of the two species. We further assess whether wild individuals can be differentiated using laboratory strains of the two species for model calibration. The NIRS spectra of C. molesta and C. pomonella differed most in the wavelengths between 1,142 and 1,338 nm. Using these wavelengths, partial least squares analysis allowed the differentiation of C. molesta and C. pomonella at the larval stage with very low error, but only as long as both the calibration and prediction sets for individuals had the same origin (either both from the laboratory or both from the field). Errors that appeared when using laboratory individuals for calibration were owing to the divergence of the C. pomonella laboratory strain, most likely following evolution during rearing. Thus, NIRS appears to be a promising tool for the easy and rapid identification of individuals in the field, provided that it is calibrated based on a subset of field individuals.