Positron emission tomography (PET) is a medical imaging technique widely used for cancer diagnostics. The objective of this study was to investigate whether dynamic 18F-FDG PET imaging could be used to characterize tumors and monitor treatment response. In this study we have looked at human prostate and breast cancer xenografts in nude mice. The first set was the androgen sensitive CWR22 prostate cancer, where the untreated tumors were compared with tumors that had received a dose of 7.5 Gy. The second set compared the basal-like MAS 98.12 with the luminal-like MAS 98.06 xenografts of breast cancer. By using different methods and techniques, a search for distinctions between groups have been conducted. It was first noticed that the very basic time activity curve (TAC) itself differed between the groups. For the latest time points, there was statistically significant differences between groups within each data set. Using a two-compartment pharmacokinetic model, rate constants describing FDG-uptake (k-parameters) have been estimated. These parameters represent some of the physiological conditions in the tumor; uptake and metabolism of glucose. For the CWR22 dataset, comparison of the mean k-parameters (treated tumors against untreated) yielded p-values of 0.06 for k1, the p-value 0.04 for k2 and the p-value 0.14 for k3. For the second dataset (MAS 98.12 against MAS 98.06) the same test yielded for k1 the p-value 0.06, for k2 the p-value 0.92 and for k3 the p-value << 0.01. By investigating every percentile of the parameter and testing, eve lower p-values could in some cases be found. By combining these parameters, an estimate of the metabolic rate of glucose (MRglc) can be obtained. The results showed that the treated CWR22 tumors were more metabolic active than the controls. We also found that the MAS 98.06 tumors had a higher metabolic rate of glucose than the MAS 98.12. Patlak plots can be used to calculate the MRglc as well and showed similar results; for the first data set a p-value << 0.01. For the second data set a p-value << 0.01 was found. This study demonstrate that both by looking at the tumor as a whole and investigating the heterogeneity it might be possible to distinguish different tumors from one another using dynamic FDG-PET.