The purpose of the study was to evaluate both CT image quality in a phantom study and feasibility in an initial case series using automated chest compression (A-CC) devices for cardiopulmonary resuscitation (CPR). Multidetector CT (MDCT) of a chest/heart phantom (Thorax-CCI, QRM, Germany) was performed with identical protocols of the phantom alone (S), the phantom together with two different A-CC devices (A: AutoPulse, Zoll, Germany; L: LUCAS, Jolife, Sweden), and the phantom with a LUCAS baseplate, but without the compression unit (L-bp). Nine radiologists evaluated image noise quantitatively (n = 244 regions, Student's t-test) and also rated image quality subjectively (1-excellent to 6-inadequate, Mann-Whitney U-test). Additionally, three patients during prolonged CPR underwent CT with A-CC devices. Mean image noise of S was increased by 1.21 using L-bp, by 3.62 using A, and by 5.94 using L (p < 0.01 each). Image quality was identical using S and L-bp (1.64 each), slightly worse with A (1.83), and significantly worse with L (2.97, p < 0.001). In all patient cases the main lesions were identified, which led to clinical key decisions. Image quality was excellent with L-bp and good with A. Under CPR conditions initial cases indicate that MDCT diagnostics supports either focused treatment or the decision to terminate efforts.