The present paper constructs a new approach named local straight-line screening (LSLS) to detect Chinese proprietary medicines (CPM) containing undeclared prescription drugs (UPD). Different from traditional methods used in analysis of multi-component spectrum, LSLS is proposed according to the characteristics of original infrared spectra of the UPD and suspected CPM, without any pattern recognition or concentration model establishment. Spectrum-subtraction leads to the variance in local straight line, which serves as a key in discrimination of whether suspected CPD is adulterated or not. Sibutramine hydrochloride, fenfluramine hydrochloride, sildenafil citrate and lovastatin were used as reference substances of UPD to analyze 16 suspected CPM samples. The results show that LSLS can obtain an accurate quantitative and qualitative analysis of suspected CPM. It is possible for the method to be potentially used in the preliminary screening of CPM containing possible UPD.