Robust partial least square regression and partial robust M-estimator regression were applied in outlier detection and quantitative analysis of nitrogen content,oil content and glucosinolate content of rapeseeds by dint of infrared photoacoustic spectroscopy.The results showed that infrared photoacoustic spectroscopy can be used in the fast determination of rapeseeds′ qualities,and that robust regressions are superior to its classic versions in the avoiding of the interference caused by outliers to optimize quantitative analysis models,and in which partial robust M-regression(PRM)performs slightly better than robust partial least squares(RPLS).