Near infrared reflectance spectroscopy (NIRS) was evaluated as a tool to predict the chemical composition of freeze-dried mutton. Samples used for the ash, dry matter (DM), crude protein (CP) and fat calibrations consisted of M. longissimus dorsi (eye muscle) from 19-month-old Merino sheep, while mineral calibrations were developed with M. semimembranosus from Merino crossbreed lambs slaughtered at a live weight of 40 kg. Samples were minced, freeze-dried and analysed according to standard laboratory procedures. Samples were scanned (1100–2500 nm) and partial least-squares regression (PLSR) was used to predict the chemical and mineral composition. Multiple correlation coefficients (r) and standard error of performance (SEP) for chemical composition constituents were: ash (0.97; 0.15%), DM (0.96; 0.38%), CP (1.00; 0.92%) and fat (1.00; 0.43%), respectively. K, P, Na, Mg, Fe and Zn showed acceptable SEP values of 600, 900, 77.89, 40, 3.15 and 3.59 mg/kg, respectively. The r values ranged from 0.86 for Zn and K to 0.92 for Mg. Very low r values (0.26–0.49) were obtained for Cu, B, Mn, Ca and Al. It was concluded that NIRS could be used as a rapid tool for predicting proximate chemical composition and certain minerals in freeze-dried mutton.