This work is to compare the kinetic parameters derived from the DCE-CT and -MR data of a group of 37 patients with cervical cancer. The modified Tofts model and the reference tissue method were applied to estimate kinetic parameters. In the MR kinetic analyses using the modified Tofts model for each patient data set, both the arterial input function (AIF) measured from DCE-MR images and a population-averaged AIF from the literature were applied to the analyses, while the measured AIF was used for the CT kinetic analysis. The kinetic parameters obtained from both modalities were compared. Significant moderate correlations were found in modified Tofts parameters [volume transfer constant(Ktrans) and rate constant (kep)] between CT and MR analysis for MR with the measured AIFs (R = 0·45, P<0·01 and R = 0·40, P<0·01 in high-Ktrans region; R = 0·38, P<0·01 and R = 0·80, P<0·01 in low-Ktrans region) as well as with the population-averaged AIF (R = 0·59, P<0·01 and R = 0·62, P<0·01 in high-Ktrans region; R = 0·50, P<0·01 and R = 0·63, P<0·01 in low-Ktrans region), respectively. In addition, from the Bland–Altman plot analysis, it was found that the systematic biases (the mean difference) between the modalities were drastically reduced in magnitude by adopting the population-averaged AIF for the MR analysis instead of the measured ones (from 51·5% to 18·9% for Ktrans and from 21·7% to 4·1% for kep in high-Ktrans region; from 73·0% to 29·4% for Ktrans and from 63·4% to 24·5% for kep in low-Ktrans region). The preliminary results showed the feasibility in the interchangeable use of the two imaging modalities in assessing cervical cancers.