This study was designed to measure organ absorbed doses from multi-detector row computed tomography (MDCT) on pediatric anthropomorphic phantoms, calculate the corresponding effective doses, and assess the influence of automatic exposure control (AEC) in terms of organ dose variations. Four anthropomorphic phantoms (phantoms represent the equivalent of a newborn, 1-, 5-, and 10-y-old child) were scanned with a four-channel MDCT coupled with a z-axis-based AEC system. Two CT torso protocols were compared: a first protocol without AEC and constant tube current-time product and a second protocol with AEC using age-adjusted noise indices. Organ absorbed doses were monitored by thermoluminescent dosimeters (LiF: Mg, Cu, P). Effective doses were calculated according to the tissue weighting factors of the International Commission on Radiological Protection (). For fixed mA acquisitions, organ doses normalized to the volume CT dose index in a 16-cm head phantom (CTDIvol16) ranged from 0.6 to 1.5 and effective doses ranged from 8.4 to 13.5 mSv. For the newborn-equivalent phantom, the AEC-modulated scan showed almost no significant dose variation compared to the fixed mA scan. For the 1-, 5- and 10-y equivalent phantoms, the use of AEC induced a significant dose decrease on chest organs (ranging from 61 to 31% for thyroid, 37 to 21% for lung, 34 to 17% for esophagus, and 39 to 10% for breast). However, AEC also induced a significant dose increase (ranging from 28 to 48% for salivary glands, 22 to 51% for bladder, and 24 to 70% for ovaries) related to the high density of skull base and pelvic bones. These dose increases should be considered before using AEC as a dose optimization tool in children.