Dose assessment in computed tomography (CT) is challenging due to the vast variety of CT scanners and imaging protocols in use. In the present study, the accurateness of a theoretical formalism implemented in the PC program CT-EXPO for dose calculation was evaluated by means of phantom measurements. Phantom measurements were performed with four 1-slice, four 4-slice and two 16-slice spiral CT scanners. Firstly, scanner-specific nCTDIw values were measured and compared with the corresponding standard values used for dose calculation. Secondly, effective doses were determined for three CT scans (head, chest and pelvis) performed at each of the ten installations from readings of thermoluminescent dosimeters distributed inside an anthropomorphic Alderson phantom and compared with the corresponding dose values computed with CT-EXPO. Differences between standard and individually measured nCTDIw values were less than 16%. Statistical analysis yielded a highly significant correlation (P<0.001) between calculated and measured effective doses. The systematic and random uncertainty of the dose values calculated using standard nCTDIw values was about -9 and ±11%, respectively. The phantom measurements and model calculations were carried out for a variety of CT scanners and representative scan protocols validate the reliability of the dosimetric formalism considered—at least for patients with a standard body size and a tube voltage of 120 kV selected for the majority of CT scans performed in our study.