Composting agro-food wastes according to industrial processes can provide organic fertilisers with known (and constant) quality levels. It is important for the user to know the degree of composting of an organic input, in order to better evaluate its potential beneficial effects on soil properties. For the industrial manufacturer, it is also important to control the quality of a compost during its elaboration. A key of success is to respect the major composting stage: the thermophilic phase. The challenge is to complete this phase without wasting any supplementary week of composting. The aim of this work was to explore the possibility of predicting by NIRS the composting degree. The parameter measured was the compost age, CA, in days. When starting a new compost pile, CA=0, and the duration of the phase is 60 days on average. Due to the heterogeneity of fresh compost samples, they were dried (40°C) and ground (<1 mm sieve) before being scanned on a NIRS 6500 (Foss NIRSystems) in ring cups. Spectra were corrected with SNVD 2,5,5 (WIN-ISI) mathematical pre-treatment and calibrations were performed using a multiple linear regression (MLR, WIN-ISI). The equation developed with 83 samples had a Standard Error of Calibration (SEC) of 9.3 d., a determination coefficient (R2) of 0.84, and a Standard Error of Cross Validation (SECV) of 9.9 d. The corresponding SD/SECV ratio (RPDcv) was 2.4. Calibration developed on a unique series of 22 samples (same pile, sampled regularly) gave a SEC of 6.0 d., a R2 of 0.96, and a SECV of 6.9. The corresponding RPDcv was 4.7. These very preliminary results show that it seems possible to assess the degree of composting, even if more efforts should be devoted to decrease the SECV to a more acceptable level (useful 4 d., ideal = 1 d.).