The purpose of this study was to determine which implementations of a T2-weighted fast spin-echo sequence of the liver resulted in observer preference in normal subjects. Five volunteers were scanned with a series of fast spin-echo sequences modified to allow for flow compensation, respiratory triggering (RT), ECG triggering, randomized phase encoding (RPE), breath-holding, and echo train length (ETL). Images were compared with conventional 2500/40/80 msec spin-echo images using flow compensation and spatial presaturation by two observers blinded to the specific sequence parameters. All FSE sequences were completed in less than the 12 minutes necessary to perform a conventional spin-echo sequence. The most preferred fast spin-echo sequence employed flow compensation, RT, and used an 8 ETL. Analysis of image preference, signal to noise, and contrast to noise showed that RT was the single most important variable in determining each image response (P < .01, P < .02, P < .01, respectively). There was some evidence that images obtained with an 8 ETL were preferred over those using a 16 ETL (P = .07). No other variables approached statistical significance although one reader preferred images with flow compensation in the frequency direction to those either not flow compensated or flow compensated in the slice direction. Respiratory triggered fast spin-echo images combined with flow compensation in the frequency direction and using ETL = 8 can provide image quality equal to conventional spin-echo sequences with significant time savings.