Temperature measurement, as one of the important parameters in flow fields, is of great significance. Toluene, a commonly used tracer in laser-induced fluorescence, is often used for flow field temperature measurement due to its temperature-sensitive fluorescence intensity. The primary reason for using toluene in temperature measurement is the redshift in its fluorescence spectrum with increasing temperature. However, there is a lack of direct analysis to elucidate the correlation between the toluene spectrum and temperature. This paper analyzes the fluorescence spectra of toluene at different temperatures using partial least squares analysis to establish the correlation between temperature and spectrum. The spectral measurement wavelength range effectively covers the entire fluorescence range of toluene. Using a 266 nm laser as the excitation wavelength for toluene, the fluorescence spectra of toluene at 35 different temperatures are measured, and the spectral measurements are accumulated 100 times to eliminate random noise. The spectra of the 35 temperature points corresponding to the 260～330 nm wavelength range are allocated to the dataset and validation set in a 6∶1 ratio. When establishing the model using the dataset, the number of factors in the model is set to 9, and the coefficient of determination of the obtained model is 0.992 2. The root mean square error of the data validation is 4.59 K. When the validation set data is applied to the model, the relative error of the predicted results is within 1%, and the root mean square error of the validation set is 1.67 K, which is in good agreement with the actual temperature values. The experimental results demonstrate the feasibility of using the partial least squares method to measure temperature through the toluene fluorescence spectrum. The establishment of this model provides new ideas and methods for the study of flow field temperature measurement. Meanwhile, the temperature results obtained from the model can also be used as calibration data for temperature measurement in flow fields where in-situ calibration is inconvenient.