The well-known process capability index CPK has been popularly used in the manufacturing industry for measuring process reproduction capability. Existing research papers related to CPK have assumed with no gauge measurement errors. Unfortunately, such assumption does not reflect real situations accurately even with highly sophisticated advanced measuring instruments. Conclusions drawn from process capability analysis are hence unreliable. In this paper, we consider estimating and testing CPK with presence of gauge measurement errors. The results show that the estimator using the sample data contaminated by the measurement errors severely underestimates the capability, resulting imperceptible power of capability testing. To obtain the true process capability, modified confidence bounds and critical values are presented to practitioners for their factory applications.