Gentamicin monitoring and the selection of the initial dosage are generally based on the relationship between pharmacokinetic parameters of gentamicin (GPP) and patient characteristics and/or clinical data (PC). However, the number of studies about this relationship in cancer patients is limited. Therefore, the main objective of the present study was to evaluate the relationship between GPP and PC in cancer patients and to identify different subgroups within this group of patients with unique relationship models between GPP and PC. A total of 198 cancer patients were included in the study. Firstly, GPP were estimated by the Sawchuk and Zaske regression method. Then, a linear regression analysis was performed to investigate the relationship between GPP and PC, and lastly subgroups with unique models were identified by comparing their regression models. The results revealed that the variable which was the best predictor of the distribution volume of gentamicine was the dosing weight (DW = IBW + (ABW-IBW), ABW being the actual body weight and IBW the ideal body weight). Creatinine clearance (CLCR) measured by a 24-hour urine collection (CLCRu) was the best predictor of gentamicin clearance (CL). When this value is not available, the CLCR estimated by the formula of Crockcroft and Gault (C-G), can be used. When the C-G formula was used, unique models to predict CL from CLCR were identified for patients who were obese, patients who had received high-dose chemotherapy and, for subjects who had never developed aplasia following chemotherapy. Whichever the model used, the results showed that some variability in pharmacokinetic parameters of gentamicin was not explained by the models, especially in some groups of patients.