Nicotine is the most important component in E cigarette liquid,whose content determines the flavor and the safety of the product.In order to improve the detecting efficiency of the nicotine content,a novel near infrared spectroscopy(NIR) combined with extreme learning machine regression(ELMR) algorithm was adopted to establish an NIR-ELMR prediction model for nicotine content in E cigarette liquid.The experimental results showed that,compared with traditional partial least squares regression(PLSR) model and principal component regression(PCR) model,the NIR-ELMR model was much better with a determination coefficient(R2) of 0.926 2,which was higher than 0.859 0 for PCR prediction model and 0.860 4 for PLSR prediction model.Besides,the root mean square error of prediction(RMSEP) for NIR-ELMR model was 0.026 8,which was smaller than 0.043 1 for PCR model and 0.040 9 for PLSR model.The above results indicated the established model could be applied to the rapid and accurate determination of the nicotine content of E cigarette liquid,which lay a foundation for the online analysis of nicotine content and the rapid determination of other quality parameters.