Fourier transform infrared spectroscopy in combination with discriminant analysis (DA) was used to classify fruit purees into three predefined groups, namely strawberry, raspberry, and apple. Using optimized parameters, 149 spectra subjected to this analysis were classified with 100% success. A separate analysis of each set of fruit was able to distinguish with 98.3% success for strawberry and 75% for raspberry, whether the fruits were fresh or freeze-thawed. Classification according to two levels of ripeness also gave good results (92.5% correctly classified) for raspberry but not for strawberry. With apple purees, DA was able to identify if sulfur dioxide had been added or not (90% correctly classified) and if the variety was Bramley or not (86% correctly classified).