PURPOSE: To investigate how a recently developed MRI based post implant dosimetric assessment technique for ultrasound guided transperineal interstitial permanent prostate brachytherapy (TIPPB) compares with the currently accepted CT based technique.MATERIALS AND METHODS: This study was based upon serial MRI and CT scans of 15 patients who had received ultrasound guided TIPPB with either I-125 (n = 8) or Pd-103 (n = 7) sources (mean number of sources 75, range 46 to 95). Each patient was imaged using both modalities within 24 hours of receiving the implant and within 1 hour of each other. Prostate volumes were determined by the same physician for both modalities. Contours and sources were digitized, and calculations performed using an in house treatment planning system with a nearest neighbor seed sorting routine and TG-43 formalism. Reference doses were 144 Gy and 115 Gy for I-125 and Pd-103, respectively. Post implant dosimetric assessment consisted of a comparison of MRI and CT based prostate volumes, dose volume histograms, and conformity and uniformity descriptors.RESULTS: The values reported are the mean ± one standard deviation for each modality. Differences (%) and ranges (%) were calculated as {(Vol M R I -Vol C T )/(Vol C T )}*100. The correlation coefficient is the Pearson product moment correlation.CONCLUSION: MRI based post implant dosimetric assessment for ultrasound guided TIPPB produces results which, while on average for the study population are similar, can be significantly different from those of CT based evaluations as is demonstrated by the weak correlations between the two groups of data. Since any assessment of implant quality is dependent upon the ability to accurately localize both the sources and target volume, differences in implant quality as determined by dosimetric results must be attributable to either source or prostate localization differences from MRI to CT, or both. The high correlation between reference isodose volumes strongly suggests that the predominant cause of the differences are due to variations in target definition. These preliminary results demonstrate that MRI based dosimetric evaluation is possible and sufficiently different from that based on CT to warrant further study. There is clearly a need to investigate whether recognition of these differences will result in improved correlation with clinical outcome.