This study has evaluated the performance of a multivariate statistical model to predict embryo implantation potential by processing data from the chemical fingerprinting of culture medium samples used for human embryo culture. The culture medium for 113 embryos from 55 patients undergoing ICSI was collected after embryo transfer. The samples were split into positive (n=29) and negative (n=84) implantation groups according their implantation outcomes (100% or 0% implantation). The samples were individually diluted and analyzed by electrospray ionization mass spectrometry (ESI-MS). The m/z ratios and relative abundances of the major ions in each spectrum were considered for partial least square discriminant analysis. Data were divided into two subsets (calibration and validation), and the models were evaluated and applied to the validation set. A total of 5987 ions were observed in the groups. The multivariate statistical model described more than 82% of the data variability. Samples of the positive group were correctly identified with 100% probability and negative samples with 70%. The culture media used for embryos that were positive or negative for successful implantation showed specific biochemical signatures that could be detected in a fast, simple, and noninvasive way by ESI-MS. To our knowledge, this is the first report that uses MS fingerprinting to predict human embryo implantation potential. This biochemical profile could help the selection of the most viable embryo, improving single-embryo transfer and thus eliminating the risk and undesirable outcomes of multiple pregnancies.