Using orthogonal partial least squares (OPLS), based on the Simca-p11.5 software, and canonical correlation analysis (CCA), performed on MatLab r2010 software, the correlation between curcuminoids extracted from Curcuma longa L. and the antitumor activity on HeLa cells was investigated to identify the significantly active constituents. Fingerprints from 31 batches of curcuminoids from C. longa L. were established using high performance liquid chromatography–electrospray ionization tandem mass spectrometry (HPLC–ESI-MS/MS), and a total of 26 selected characteristic peaks were quantitatively analyzed. Afterward, the antitumor activities of the curcuminoids on HeLa cells were measured using an MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay. We found that 13 of the curcuminoids (peaks 9, 18, 14, 8, 16, 17, 24, 12, 4, 13, 10, 20 and 11) were significantly correlated with antitumor activity via a Loadings plot and VIP (variable importance in projection) in OPLS and a correlation coefficient in CCA. These results support a method for the discovery of antitumor active constituents.