‘Granny Smith’ apples are highly susceptible to skin browning (=scald). These disorders can significantly reduce fruit’s marketability due to their unattractive appearance. Superficial scald and sunscald have different etiologies but their symptoms are similar and easily mistaken. The oxidation of α-farnesene causes superficial scald (SS); on the other hand, sunscald (SC) is a non-oxidative process that affects only the sunlit sides of the fruit. To correlate the spectral fingerprint with fruit susceptibility to these disorders, hyperspectral images (400-1000 nm, 640x865 px; Headwall Photonics, Bolton, MA) were taken at harvest from sun-exposed and unexposed sides of the fruit (n=216; ~26,000 px) and later, after six months in air storage (33 oF), correlated with scald incidence. The dataset grouped 145 asymptomatic, 117 with SS and 170 with SC fruit. After pre-processing spectral information (Savitzky-Golay dev, standard normal variate), iPLS wavelength selection showed that bands from 400 to 650 and 900 to 950 nm were the most accurate for at-harvest spectral differentiation between superficial and sunscald symptoms. A neural network classification model was trained (18,226 px) and validated on an independent dataset (7,808 px), achieving overall accuracies of 78 % and 73 %, respectively. After prediction, SC px obtained the highest classification metrics (accuracy 87 %, precision 86 %); meanwhile, the asymptomatic class showed the lowest metrics (accuracy 74 %, precision 60 %). These preliminary results showed that in the same way, sunscald could be identified close to harvest using hyperspectral fingerprints, superficial scald could also be predicted at harvest on susceptible fruit and differentiated from sunscald susceptible fruit.