This study demonstrates methodology for predicting the quality (in terms, for instance, of specific activities) of recombinant proteins expressed by mammalian cell cultures, at upstream process stages, using metabolic fingerprint analysis. Metabolites extracted from samples taken from a culture of a transfected Chinese Hamster Ovary (CHO) cell line expressing a recombinant protein in a bioreactor at various days of the cell culture process were analyzed by ultra high-pressure liquid chromatography-electrospray ionization-time of flight mass spectrometry (UHPLC-ESI-TOFMS). The LC-MS data were processed and the extracted information was correlated with the concentration of the active protein by partial least squares (PLS) regression, which revealed strong correlations between the LC-MS results and the concentration of active protein (R
2 = 0.99). The correlations between the LC-MS data and other parameters (glucose concentration, lactate concentration and the number of viable cells) were also studied. To obtain an overview of the data, Principal Component Analysis (PCA) was applied to the LC-MS data obtained from the samples to observe clustering or separation in the sample set. The PCA indicated that the LC-MS data obtained from samples from different days were significantly separated in temporal order from day 7 to day 28, according to the first Principal Component.