Silica-based monolithic column material was synthesized and an enrichment device was fabricated with the material by assembling the material inside a glass column.The enrichment device was applied for the determination of micro-carbaryl with near-infrared spectroscopy(NIRS).The aqueous solutions of carbaryl passed through the device and the carbaryl was enriched on the surface of the material where diffuse reflection NIR spectra were measured.These procedures of enrichment and measurement ensured to concentrate analytes for the measurement,so that the sensitivity of determination of NIRS could be improved.NIR spectra of carbaryl solutions(0.01-1.00μg mL~(-1)),measured after the application of the enrichment device,were pretreated with multiplicative scatter correction(MSC) and regressed against the concentrations of the carbaryl solutions with partial least squares (PLS) method.The results showed that the minimum value of root-mean-square error of prediction(RMSEP) was 0.1771μg mL~(-1) when the number of latent variables was 3 of PLS regression.Therefore,the number of latent variables 3 was selected as the optimum value.RMSEP was not very low but acceptable considering that NIRS is commonly used in macro amount analysis and it is quite difficult for NIRS to determine micro amount analytes,especially,less than 1μg mL~(-1) Silica-based monolithic column material was synthesized and an enrichment device was fabricated with the material by assembling the material inside a glass column. Enrichment device was applied for the determination of micro-carbaryl with near-infrared spectroscopy (NIRS). The aqueous solutions of carbaryl passed through the device and the carbaryl was enriched on the surface of the material where diffuse reflection NIR spectra were measured. These procedures of enrichment and measurement of to concentrate analytes for the measurement, so that the sensitivity of determination of NIRS could be improved .NIR spectra of carbaryl solutions (0.01-1.00 μg mL -1), measured after the application of the enrichment device, were pretreated with multiplicative scatter correction (MSC) and regressed against the concentrations of the carbaryl solutions with partial least squares (PLS) method. The results showed that the minimum value of root-mean-square error of prediction (RMSEP) was 0.1771 μg mL ~ (- 1) when the number of latent variables was 3 of PLS ​​regression.Therefore, the number of latent variables 3 was selected as the optimum value. RMSEP was not very low but likely to consider NIRS is commonly used in macro amount analysis and it is quite difficult for NIRS to determine micro amount analytes, especially, less than 1 μg mL ~ (-1)