This paper discusses the various uses of bulk-method-determined fluxes and the inherent difficulties of making flux measurements over the ocean. The accuracy of eddy correlation, profile, dissipation, and bulk flux methods is compared. We conclude that there are four principal sources of uncertainty in bulk flux determinations made using ship data: the accuracy of the flux measurements used to develop the bulk method, the variations resulting from the use of different bulk coefficient schemes, the accuracy of the meteorological sensors, and the distortion of the measurements produced by the ship. This paper also describes previously published analyses of each of these uncertainty sources and compares the relative importance of those sources. Finally, the results are combined to estimate the overall accuracy of the method. Our analysis determined that the average accuracy of the bulk method ranges from 35% to 105% for stress magnitudes of 0.025--1.0 N/m2, 35--220% for sensible heat flux magnitudes of 5--150 W/m2, 40--215% for latent heat flux magnitudes of 10--300 W/m2, 85% to >1000% for Monin-Obukhov stability magnitudes at 10 m of 0.001--10, and 160% to >1¿104% for roughness length magnitudes of 10-5--10-2 m. The results of the analysis are presented graphically and in tabular and polynomial forms to facilitate their use by those wishing to test the sensitivity of their work to the typical average uncertainty of bulk method flux, stability, and sea surface roughness determinations. ¿American Geophysical Union 1987