This research addresses the problem of predicting white specks in dyed cotton fabrics. The main objective is to predict fabric quality from bale fiber properties given controlled mill processing. Mill processing must be controlled because past studies show mechanical processing can cause major differences in white speck levels. The fabrics made from the US 2001 ATMI (American Textile Manufacturers Institute) Cotton Variety Processing Trials were spun at CQRS on full scale equipment in bale quantities and analyzed using Autorate-2-03 (AR-2-03), to develop the initial prediction equations using fiber data from high-speed measurement systems. The recent release of Autorate in May 2005 (AR-5-05) was developed by Dr. Bugao Xu to measure white specks on dark fabrics in conjunction with this research, and removes operator input. These results are compared to the white speck measured by AR-2-03 and compared to the fiber data generated by AFISPRO. The study looks at relationships to AFISPRO data at different stages in mill processing. Fabrics produced from bale quantities at CQRS are compared to 20 lb lots at SRRC. White speck levels also change when the same yarns are woven on the shuttle loom vs. the rapier loom. The information gathered has helped set a long term protocol for white speck processing research at the Southern Regional Research Center and is the beginning of a large data base which will include international cottons to develop the final WSP (White Speck Potential) equations. This information will be useful to as a tool to measure the effects of field and ginning practices on the levels of white specks without having to carry the research out to finished fabrics. Cotton breeders will be able to use the equations in the development of new varieties with low white speck potential, by eliminating varieties with high white speck potential early on.