Cumulative count of conforming (CCC-r) charts are usually used to monitor non-conforming fraction p in high-yield processes. Existing approaches to setting the control limits may cause non-maximal or biased in-control average run length (ARL). Non-maximal in-control ARL implies that the chart might not quickly detect the upward shift of p from its nominal value p0. On the other hand, biased in-control ARL means that both the in-control and out-of-control ARLs are inflated. This paper develops a new approach to setting control limits for CCC-r charts with near-maximal and near-unbiased in-control ARL. Experimental results show that the proposed approach is effective in terms of the maximization and unbiasedness of in-control ARL. Copyright © 2009 John Wiley & Sons, Ltd.