Context: Meningiomas account for 30% of primary brain tumors. A dramatic increasing of their incidence with an annual percentage change ranging from 2 to 4.5% has recently been highlighted by several sources including either symptomatic or asymptomatic tumor. In case of asymptomatic lesions the issue of the frequency of control is a major concern for public health and neurosurgery. The aim of this work is to validate clinically a tumor growth model that can predict volume and shape of the meningioma at a later time. Material and methods: The mathematical model we present offers a mechanistic description of the healthy and tumor cell densities evolution over time in function of patient-specific parameters computed from the tumor volume variation between two times. These volumes are determined from T1 gadolinium injected MRI thanks to home-designed semi-automatic segmentation software. Once the model is personalized, we predict the tumor volume at a later time and the spatial growth of the tumor, which is compared to the 3D segmentation at a later time thanks to the DICE coefficient (two times the ratio of the volumes intersection with their union). Patients: Patients included in the project present asymptomatic meningiomas with various locations. Histology confirmed meningiomas for 7 patients who benefits of a surgical resection. We gather at least three MRI examinations per patients with an average time interval of 1 year between two imaging procedure. A primary cohort of 8 patients (7 women / 1 men, median age: 49 years, standard deviation: 11.7 years) was used to design the mathematical model in order to reproduce the meningioma evolution in terms of volume and 3D spatial extension. The second cohort of 30 patients (27 w / 3 m, med: 52 y, stdev: 13.2 y) was used to validate the model by its personalization with the two first MRI and the prediction of the tumor volume at the time of the third exam. Finally, a prospective cohort of 18 patients (18 w, med: 52 y, stdev: 13.6 y), where the third MRI was not available at the time of the prediction, has been used to evaluate the method in clinical conditions. Results: The study of the first cohort allowed us to validate the 3D tumor growth simulation method where we found an excellent agreement with a DICE value of 84%. The validation of the prediction method on the second cohort results on a reliable accuracy within a ±15% relative error between the predicted and observed volumes. This result has been confirmed thanks to the prospective cohort for which we found an average relative error of 12% on the tumor volume prediction. Conclusions: The accuracy of these predictions led us to develop a medical software allowing clinicians to predict the tumor growth of meningiomas directly from their MRI data and optimize the frequency of MRI.