The aim of this study was to evaluate the potential of Curie point pyrolysis–mass spectrometry (CpPy–MS) to estimate the collagen and lipids contents and the texture of lamb meat. The work was carried out using the longissimus muscle from 120 commercial lambs originating from different European countries and from contrasting production systems. Meat characteristics varied largely between animals, e.g. 582 (S.D. ± 124) μg hydroxyproline g −1 for collagen content and 22.5 (S.D. ± 9.6) mg g −1 for lipid content, and 2.9 (S.D. ± 1.2) kg cm −2 for shear force value. Non-linear models of prediction were calculated by Artificial Neural Networks (ANNs). The results showed that it is possible to evaluate lipid and collagen contents and meat texture from CpPy–MS fingerprints. The best predictions obtained with test samples were for lipids ( r = 0.89; mean prediction error = 10%), for collagen ( r = 0.85; mean prediction error = 11%) and for texture ( r = 0.78; mean prediction error = 12%).