The quantitative models of cholesterol and triglyceride analysis in whole blood are established by using near-infrared transmission spectroscopy combined with chemo-metrics methods. First the transmission spectra of whole blood in wavelength range of near infrared full spectrum (800~2500 nm) is acquired, and then the calibration models for cholesterol and triglyceride analysis are optimized by using interval partial least squares (iPLS). The results show that the optimal wave band of calibration models for cholesterol and triglyceride lies 1650~1730 nm and 2260~2340 nm respectively, and the prediction coefficient of optimal analysis model is 0.79 and 0.865 respectively, the root mean square errors of prediction of related prediction model are 0.5 mol/kL and 0.28 mol/kL respectively. The study demonstrates that compared with whole spectra data model, the iPLS model can not only improve precision, but also simplify the model. Near-infrared spectroscopy combined with iPLS can be used successfully to determine the content of cholesterol and triglyceride in whole blood.