Dynamics of tumor growth of a given type of cancer are very different from one patient to another. Numerous parameters can affect such growth. On the one hand, genetic mutations may be responsible for the activation or inactivation of certain cell signaling pathways and, on the other hand, host’s response to tumor growth crucially depends on patient’s general state. Predicting the evolution of a tumor is rather difficult for clinician. The nonlinear dynamical systems theory is very helpful to investigate and understand complex phenomena. Its use in oncology could thus offer interesting prospects for the individualization of treatments. We started by constructing a mathematical model to reproduce the evolution of a prostatic adenocarcinoma after radical prostatectomy. Thanks to a genetic algorithm, we obtained a few sets of parameters allowing us to accurately reproduce changes in the PSA level measured in some patients. These models allowed us to study the impact of different intermittent therapies by GnRH analogs. Then, we developed a mathematical model to reproduce the evolution of tumor growth in a two-dimensional domain and investigate the impact of different sets of parameter reproducing certain therapies or the influence of vascular geometry on the form of a pulmonary squamous cell carcinoma. This work represents a first approach towards the individualization of tumor growth and will ultimately be used clinically for optimizing therapeutics, reducing recurrences and improving survival.