A study developed to test the hypothesis of a possible relationship between metabolic modifications and chromosomal imbalances in solid tumors leads us to investigate the metabolism of purine nucleotides in human gliomas. In order to assess the representativeness of experimental models frequently used, the activities of nine enzymes involved in the synthesis and in the catabolism of purine nucleotides were measured on samples of normal brain, primary and xenografted glial tumors and cell cultures established from human gliomas. In parallel, two enzymes involved in pyrimidine metabolism were also studied on the same samples. The results highlight the low activity of the purine metabolism in human gliomas when compared to normal brain, tissue with low proliferative activity. On the contrary, the pyrimidine metabolism in human gliomas is increased by comparison to normal brain. For the purine metabolism, few differences are observed between enzyme activities calculated in primary glial tumors, xenografts and cells in culture. In grafted tumors and cell cultures, the activity of this metabolism is similar or lower than in normal brain, except for inosine monophosphate dehydrogenase. However, for the pyrimidine metabolism, significantly differences are observed between primary glial tumors, grafted glial tumors and cell cultures. The thymidine kinase/thymidylate synthetase ratio depends on the model studied. These results point out the problem of the representativeness of these models, especially when used for experimental therapeutic studies. This metabolic study also underlines that all results should be interpreted carefully and that the limits for the use of these two experimental models should always be clearly exposed.