Tool wear is an inevitable problem in many machining processes. The problem constitutes an inseparable component of variation and can be considered as a systematic assignable cause of process variability. Tool replacement should be initiated when the yield drops below a certain level. Determining the best time for tool replacement is essential to balance between production quality and tool utilization. The yield index, Spk, has been effectively applied to assess the yield of processes. However, its ordinary measure is inaccurate when data is contaminated by inseparable nonrandom variation.Yield evaluation becomes imprecise when a process is subjected to the tool wear problem. Thus, this paper presents an efficient procedure that determines the best time to replace tools under an extremely low fraction of defectives. The variation from assignable causes is removed by using the linear regression technique. A modified estimator of the Spk index with its distribution is proposed to evaluate the processes with the tool wear problem .