Infrared spectroscopy is known as a rapid tool to study apricot fruit quality, non-destructively by near-infrared ( NIR, 800-2500 nm) range and destructively on fruit puree by mid-infrared ( MIR, 4000-650 cm(-1)) range. These techniques appear suitable for the determination of soluble solids content ( SSC) and titratable acidity ( TA). However, a major drawback is the need to repeatedly establish calibration equations, which can depend on cultivar, year, etc. The objective here was to evaluate the robustness of NIR and MIR techniques and to determine the best strategy: is it more efficient to build models every year or to build global models combining all years? Eight apricot cultivars, representative of the known apricot diversity for quality traits, have been analysed between 2005 and 2011. The most efficient strategy ( i. e., reliable calibration with a minimum of calibration) was a multi-year model, integrating a limited amount of fruit per year but incremented each year.