This paper presents an automatic method for segmenting pediatric brain MR images without using atlas priors. The proposed approach comprises of skull stripping, extraction of the cerebrospinal fluid, removal of extraneous portions and final brain segmentation. The method is tested on pediatric images in the age range 1 to 1430 days and the results are evaluated both qualitatively and quantitatively. The algorithm achieves an average Dice value of 0.9784, Jaccard index of 0.9578, sensitivity of 0.9750 and specificity of 0.9898. © 2016 IEEE.