Near-infrared reflectance spectroscopy (NIRS) technology was used for the non-destructive measurement of melon-pulp colour (a*, b*, C* and h*), one of the main indicators of ripeness and quality. A total of 432 Cantaloupe and Galia melons were used in the construction of calibration models, testing various spectral signal pretreatments and both linear and non-linear regression algorithms. The coefficient of determination (r2) and the standard error of cross-validation (SECV) obtained for parameters a* (0.96, 2.16), b* (0.85, 3.25), C* (0.82, 3.76) and h* (0.96, 3.64) in intact fruit confirmed the a priori viability of NIRS technology with modified partial least squares (MPLS) regression for measuring melon ripeness and quality. Moreover, the application of a local algorithm improved the ability of models to predict all the internal-colour quality parameters studied. These results suggest that NIRS technology is a promising tool for monitoring ripening in melons and thus for establishing the optimal harvesting time.