Portable near-infrared (NIR) spectrometer for quick on-site measurement is an important trend in the study field of NIR spectroscopy. However, in order to achieve quick measurement, a portable NIR spectrometer is generally not equipped with temperature-controlled device. Therefore, the change in ambient temperature will bring a relatively large measurement error to the predicted results. Reducing the error caused by the changes in ambient temperature is an important problem that has to be solved before the large-scale application of portable near-infrared spectrometer in the field of quick on-site measurement. The condensation points of diesel is an important parameter to evaluate diesel quality and temperature range for diesel application. Development of the on-site quick measurement of condensation point can effectively reduce the cost of traditional measurement. In the present study, the NIR spectra are collected by a portable spectrometer in the wavelength range of 950~1 650 nm for 50 kinds of diesel samples with different condensation points. The effect of changes in ambient temperature on the quantitative analysis results is studied using one new type of NIR spectrometer. This type of spectrometer is a portable spectrometer designed based on a digital micromirror device(DMD), which is developed for quick on-site measurement without temperature-controlled device sample cell. Firstly, the predicting model is developed for condensation point under the condition of ambient temperature at T₀=25 ℃, using based on the partial least square method. Then, the spectra measured under other ambient temperatures (T_E=-10, 0, 10, 20, 30, 40 and 50 ℃) are introduced into this model to predict the condensation point, and the relationship between prediction error and changes in ambient temperature (T_E-T₀) is studied. The linear function fitted the relationship between prediction error and ambient temperature. It is found that the average value of condensation point prediction error is Δc=-0.019 8(T_E-T₀). The compensation factor of environmental temperature is brought into the prediction model developed under 25 ℃, and a temperature compensation model for the change in ambient temperature is established to predict the condensation point of diesel with NIR spectra collected under other conditions ambient temperature. The root means square error (RMSE) of condensation point prediction at 10 ℃ is improved from 14.6 to 8.8, and the coefficient of determination increased from 0.4 to 0.7. The study shows that the temperature compensation model can effectively reduce the error caused by ambient temperature. This method can improve the time cost for developingthe model and extend the temperature range in applying a portable NIR spectrometer.