An experiment was conducted to remove seed coat fragments (SCF) with a saw-type lint cleaner using newly designed grid bars. The test consisted of one control and five experimental grid bar designs, and a treatment that contained no lint cleaning (No LC). Two types of cotton were used, a common upland cultivar and a cultivar known to have a fragile seed coat that breaks easily and contaminates lint with SCF. Results showed that fiber from the Acala cultivar was longer, more uniform, contained less short fiber, had fewer neps, and fewer seed coat neps (SCN) than the fragile cultivar. There were differences in fiber properties among lint cleaning treatments, but those differences were between the treatment that contained No LC and the grid bar treatments collectively. Visible foreign matter in ginned lint was not different among lint cleaning treatments. Both AFIS seed coat nep count and a manual count of SCF, which were used as indicators for SCF, were not different among lint cleaning treatments. As expected, fiber in the lint cleaner waste was shorter, and contained more short fiber and more neps than fiber in the bale, but SCN count in the lint cleaner waste also was not different among grid bar designs. High-speed videography showed that SCF were not removed by colliding with the grid bars, but were actually drawn back into the lint stream by attached fiber. Other means to separate SCF from ginned fiber are being investigated.