In a recent Beltwide Conference (2005), the author reported that the proper choice of mechanical sample-blending and specimen-compositing approaches could reduce substantially the variation in length properties measured by the Advanced Fiber Information System (AFIS) on raw cotton lint. The improvement allowed measurements on raw lint to approximate the reproducibility obtainable from high-quality textile sliver. So to obtain the most precise length measurements from limited specimens of cotton lint by AFIS, a mechanical blending methodology is required. This precision methodology was used herein to study the effect of mass feed-rate on fiber properties measured by the AFIS. Hand sliver from mechanically blended fiber was fed to the instrument at a constant feed-roll speed. Each of eight cottons was tested at five levels of sliver linear density, replicated by two operators. Each test was replicated internally by the AFIS five times at 5,000 fibers per replicate, for a total of 25,000 fibers per specimen. In addition to the fiber properties normally reported by AFIS, the length distributions were used to calculate the Broken Fiber Content (BFC). Analysis was performed on the AFIS properties and BFC as a function of fiber loading on the opening and sensor stages of the AFIS. This analysis helped to illuminate the degree to which fiber loading on the AFIS contributed to downstream variations in sensed values of the measured fiber properties. It was found that the proper combination of gentle mechanical blending and less aggressive loading of the AFIS opening roll gave the most efficient testing, i.e., the most precise results with the least measurement overhead. This information should be of particular interest to breeders, who have large numbers of small samples for screening.