A modified polyacrylonitrile (PAN) hollow-fiber membrane from a commercial source has been applied as the separation channel in flow field-flow fractionation (FFF). With the PAN membrane fiber, the application range of flow FFF could be extended to synthetic polymers that are soluble in a variety of organic solvents. The PAN membrane was shown to be resistant to hydrophobic solvents, such as dichloromethane (DCM), tetrahydrofuran (THF), ethyl acetate, and methyl ethyl ketone (MEK), as was illustrated by the successful fractionation of different polymer standards in these solvents. The system performance was assessed using polystyrene (PS) standards with ethyl acetate as the solvent. For a 100 kDa PS standard, the average recoverywas 57%, but for standards with a molar mass of 400 kDa and higher, 100% recovery was obtained. A linear relationship between peak area and injected mass was found. The run-to-run and fiber-to-fiber repeatability was determined using 100- and 400 kDa PS standards. The repeatability appeared to be satisfactory, with relative standard deviations < 2% for the retention times and < 5% for the recoveries of the standards. Plate numbers for the 400 kDa standard on different fibers were in the order of 110. From measurements on the fractionation of ferritin aggregates, it is concluded that the instrumental band-broadening is negligible. For an accurate determination of diffusion coefficients and molecular sizes based on retention times, calibration of the channel with standards appeared to be necessary. However, it was shown that the FFF system could be coupled to a multiangle light scattering (MALS) detector, thus providing an alternative on-line method for calibration. Expressions for the maximum obtainable plate number per unit of time have been derived for a hollow-fiber flow FFF system. It is shown that an increase in the system performance can be expected from a scaling down of the fiber diameter.