Objectives: Temperature dependent chemical shifts of important brain metabolites measured by localised 1H MRS were investigated to test how the use of incorrect prior knowledge on chemical shifts impairs the quantication of metabolite concentrations. Materials and methods: Phantom measurements on solutions containing 11 metabolites were performed on a 7 T scanner between 1 and 43 °C. The temperature dependence of the chemical shift differences was fitted by a linear model. Spectra were simulated for di erent temperatures and analysed by the AQSES program (jMRUI 5.2) using model functions with chemical shift values for 37 °C. Results: Large differences in the temperature dependence of the chemical shift differences were determined with a maximum slope of about ±7.5 × 10−4 ppm/K. For 32–40 °C, only minor quantification errors resulted from using incorrect chemical shifts, with the exception of Cr and PCr. For 1–10 °C considerable quantification errors occurred if the temperature dependence of the chemical shifts was neglected. Conclusion: If 1H MRS measurements are not performed at 37 °C, for which the published chemical shift values have been determined, the temperature dependence of chemical shifts should be considered to avoid systematic quantification errors, particularly for measurements on animal models at lower temperatures.