在采用近红外漫反射光谱对浑浊介质中的物质浓度做定量分析时，物质浓度测量的灵敏度与所选的光源-检测器位置有较大的关联。解析了物质浓度测量的灵敏度随光源-检测器距离变化的曲线特点，给出了三个特殊位置：灵敏度全局极大位置、灵敏度为零的位置和灵敏度局部极大的位置。以葡萄糖在某浑浊介质溶液（含血红蛋白、白蛋白的Intralipid溶液）中的测量为例，从扩散方程、蒙特卡罗模拟、漫反射光谱测量实验三方面，均验证了灵敏度曲线中存在的这三个特殊位置。另一方面，考虑到近红外光源的稳定性、铟镓砷检测器的量子效率与系统噪声等因素的限制，采用仪器的信噪比作为另一个客观指标，对实际测量可得到的灵敏度进行了评价。综合灵敏度曲线的特点和仪器的实际测量水平，给出了测量葡萄糖的最佳位置，可实现最优的测量灵敏度与测量精度。采用优化后的测量位置，葡萄糖的测量灵敏度可提高10倍以上。测量葡萄糖灵敏度的研究对人体组织中的无创血糖检测有较好的参考价值，同时，对漫射光谱灵敏度的解析方法可被借鉴于其他物质的近红外漫反射光谱测量位置的优选。

For measurement of substance concentration in turbid medium by near-infrared diffuse spectroscopy, the measurement sensitivity and the position of the adopted spectrum detector, i.e. source-detector separation (SDS), are closely related. By analyzing the sensitivity of substance concentration measurement curved with changes of SDS, three special SDSs with maximum, minimum (zero) and local maximum sensitivities were obtained, respectively. In the glucose measurement experiment in turbid medium solution, Intralipid solution with water, glucose, hemoglobin and albumin, the three special positions of SDS were verified by diffusion equation, Monte Carlo simulation and diffuse reflection spectroscopy. Considering the limiting factors such as stability of light source, detective quantum efficiency, noises of InGaAs detector, we used the signal-noise ratio (SNR) as another indicator to evaluate the sensibility of an actual measurement. After comprehensively analyzing the sensibility graphs and actual measurement conditions, we presented the optimized SDS which the glucose measurement precision was improved by more than 10 times. The method for improving glucose measurement sensitivity in turbid medium could be referenced and practiced in non-invasive blood glucose measurement or other substance concentration measurement using near-infrared spectroscopy.