The purpose of this study was to evaluate the consequences of different choices of acquisition parameters on the actual image noise and on the patient dose with an automatic tube current modulation system. The CT investigated was a GE Lightspeed 16-slice and an anthropomorphic phantom was used to simulate the patient. Several acquisitions were made varying noise index (NI), kilovoltage and pitch values. Tube current values were compared for the different acquisitions. Patient dose was evaluated in terms of volumetric computed tomography dose index (CTDI(vol)) and also as effective dose. The noise actually present in the images was analyzed by a region of interest analysis considering representatively phantom sections in the regions of the shoulders, of the lungs and of the abdomen. The obtained results generally evidenced a good agreement between the noise index and the measured noise for the abdomen sections, whereas for the shoulders and the lungs sections the measured noise was respectively greater and lower of the NI. Varying the kV the automatic current modulation system provided images with a substantially constancy of the actual noise and of the patient dose. An increase of the pitch generally decreased the patient dose, whereas the noise was slightly greater for the lowest pitch and almost constant for the other pitch values. This study outlines some important relationships between an automatic tube current modulation system and other CT acquisition parameters, providing useful informations for the choice requested by radiologists in the task of optimization of the CT acquisition protocols. Unless there are other considerations in place, pixel pitches below 1.375 should be avoided, and kVp settings can be changed with no real impact on dose or image noise.