Mechanoluminescence (ML) is used in many areas, in particular for ML-based optical thermometry. However, ML-based thermometry is limited owing to the lack of a strong theoretical foundation. Here, using CaZnOS:Er3+ and the well-established Boltzmann distribution, a solid and novel ML thermometry framework is established. Upon external force stimulation, CaZnOS:Er3+ generates bright green ML ascribed to the 2H11/2 → 4I15/2 and 4S3/2 → 4I15/2 transitions of Er3+. And the populations at the 2H11/2 and 4S3/2 excited states are confirmed to follow the Boltzmann distribution during the ML process. The utility of this novel ML thermometry framework is demonstrated here on a practical example. Further, potential applications of CaZnOS:Er3+ to anti-counterfeiting and information encryption are discussed. These results demonstrate that CaZnOS:Er3+ luminescent materials are promising for multifunctional applications (especially for temperature sensing), and provide the foundation for future ML applications.