PURPOSE: To compare the efficacy of breathhold and non-breathhold sequences in the detection of hepatocellular carcinoma (HCC) in cirrhotic patients using superparamagnetic iron oxide (SPIO)-enhanced MR imaging, and to determine the optimal sequence combination. MATERIALS AND METHODS: By means of unenhanced and iron-oxide-enhanced MRI, 29 patients with 49 nodular HCCs were evaluated for the presence of HCC nodules. Twenty-one were male and eight were female, and their ages ranged from 38 to 71 (mean, 56) years. Eight different MR sequences were used, including four non-breath-hold sequences and four breath-hold, and images were obtained before and after the administration of S-PIO particles. Non-breath-hold sequences included T2-, proton density-weighted SE, and TSE imaging, while breath-hold sequences comprised T1-weighted fast low-angle shot (T1wFLASH), half-Fourier acquisition single shot turbo spine echo (HASTE), T2-weighted fast imaging with steady-state free precession (T2*wFISP) and T2-weighted breath-hold TSE (T2wBHTSE). Image analysis involved both quantitative and qualitative analysis. The quantitative parameters calculated were signal-to-noise (S/N) ratios for livers and tumors, contrast to noise (C/N) ratios for tumors seen on precontrast and postcontrast images, and percentage of signal intensity loss (P-SIL) after SPIO injection. Images were analysed qualitatively in terms of image artifacts and lesion conspicuity, and prior to calculating sensitivity, the numbers of lesions detected using various pulse sequences were counted. RESULTS: SPIO had a marked effect on liver S/N ratio but a minimal effect on tumor S/N ratio. PSIL was best in T2*wFISP images, while T2wSE images showed the second-best results (p< 0.05). Tumor-to-liver C/N values were also highest with T2*wFISP, while T2wTSE and HASTE images were next. Qualitative study showed that non-breath hold images and FISP were better than breath hold images in terms of lesion conspicuity. The latter, however, were much better than non-breath-hold images with regard to image artifacts (p <0 . 0 5 ) . Sensitivity after the injection of contrast material increased in every image sequence except T1wFLASH, and,in particular, postcontrast FISP and T2wTSE showed the best results (p> 0.05). CONCLUSION: SPIO-enhanced MR imaging effectively detected hepatocellular carcinoma in cirrhotic livers. In terms of lesion detection and improvement of the lesion to liver C/N ratio, the FISP sequence was at least as good as non-breath-hold sequences, but if the T2 suppression effect of SPIO is to be obtained, other breath-hold sequences are not appropriate. To help lesion characterization, we suggest that T1w-FLASH and non-breath- hold T2w-TSE imaging are added to the optimal SPIO-enhanced MR imaging sequence.