PURPOSE: To construct a temperature-controlled diffusion phantom with known diffusion properties and geometry in order to facilitate the comparison and optimization of diffusion sequences with the objective of increasing the precision of experimentally derived diffusion parameters. MATERIALS AND METHODS: A temperature-stabilized diffusion phantom made up of two crossing strands of hydrophobic polyethylene fibers was constructed. Reproducibility and temperature dependence of several diffusion parameters was investigated and compared with computer simulations. Furthermore, in order to stimulate actual use, the precision of measurement of different diffusion-encoding schemes was compared using bootstrap analysis. RESULTS: The measured values of the diffusion parameters are highly reproducible and feature strong temperature dependence which is reproduced in simulations, underlining the necessity of a temperature-stabilized environment for quality control. The exemplary application presented here demonstrates that the phantom allows comparing and optimizing different diffusion sequences with regard to their measurement precision. CONCLUSION: The present work demonstrates that the diffusion phantom facilitates and improves the comparison and quality control of diffusion sequences and the ensuing parameters. The results show that an accurate temperature control is a vital prerequisite for highly reproducible calibration measurements. As such, the phantom might provide a valuable calibration tool for clinical studies.