Our purpose was to investigate whether in vivo proton magnetic resonance spectroscopic imaging, using normalized concentrations of total choline (tCho) and total creatine (tCr), can differentiate between WHO grade I pilocytic astrocytoma (PA) and diffuse, fibrillary WHO grade II astrocytoma (DA) in children. Data from 16 children with astrocytomas (11 children with PA and 5 children with DA) were evaluated retrospectively. MRS was performed before treatment in all patients with histologically proven low-grade astrocytomas. Metabolite concentrations of tCho and tCr were normalized to the respective concentration in contralateral brain tissue. The Mann-Whitney U test was performed to evaluate differences between these two groups. Normalized tCho did not show any statistically significant difference between the two groups. There was a strong trend (P = 0.07) toward higher values of normalized tCr in the DA group. For 3 of 5 children with DA, lactate was detectable, but only 1 of 11 children with PA showed lactate. We concluded that choline as a single parameter is not reliable in the differential diagnosis of low-grade astrocytomas in children. Our results suggest that tCr concentrations combined with lactate will be helpful in the differential diagnosis of PA and DA in children.