A chemometric model for screening chemical hazards in infant formula has been built based on liquid chromatography coupled to quadrupole/time of flight (Triple TOF®) mass spectrometry. Six chemicals, naming as salbutamol, lincomycin, sulfadiazine, spiramycin and megestrol acetate, were deliberately selected as the representatives of unknown chemical hazards to be spiked into infant formula. Fifteen different infant formula samples, with equal amount, were dissolved into water and mixed homogeneously to prepare test samples representing typical infant formula matrices. All test samples were divided into reference group (free from spiked chemicals) and spiked group. The test samples were extracted with acetonitrile, and then cleaned-up with Captiva NDLipids filtering solid phase extraction column. The liquid chromatographic separation was made on a CORTECSTM C18+column using mobile phases of 0.3%(v/v) formic acid-5%(v/v) water in acetonitrile and 0.3%(v/v) formic acid-5%(v/v) acetonitrile in water with gradient elution. Mass spectrometric analysis was conducted by a mixed acquisition mode of TOF full mass scan-information dependent acquisition (IDA)-product ion scan. Acquired data were exported to MarkerViewTM software (version 1.2.1.1, SCIEX, USA) for peak finding, alignment and filtering. Then the preprocessed data were subject to SIMCA-P software (version 14.0, Umetrics AB, Sweden) for discriminant analysis. An orthogonal partial least squares-discriminant analysis (OPLS-DA) model was built with good fitness and good prediction ability (R2X(cum)=0.742, R2Y(cum)=0.997, Q2Y(cum)=0.905). The reference and spiked samples were well grouped and the six chemicals (spiking level down to 50 μg/kg, based on weight of infant formula) were reliably identified by S-Plot of OPLS-DA model. The technological system for monitoring quality and safety of infant formula could be benefited from the proposed approach.