This thesis presents new techniques to investigate and understand the source of process variability in plasma etching. In a semiconductor factory thousands of wafers are processed every month in multiple chambers. Whi le great effort is made to create reproducible process conditions, common and special cause variation remain a big challenge for the semiconductor industry. Process conditions are never identical from wafer to wafer and chamber to chamber. When high-frequency RF power, employed to create a plasma, is coupled into a chamber, the electrical characteristics of each chamber assembly is different. This electrical difference is as a result of mechanical differences of chamber components and how they are assembled. RF losses of the current affect the power deposition in the plasma and affect the process outcome. As each chamber processes more and more wafers, by-products buildup on the chamber walls impacting the process repeatability and influencing the processing chemistry. The surface roughness of the electrode and other chamber materials impact the rate at which the by-products deposit, which may also affect the process repeatability both chemically and electrically. These sources of variation contribute to inconsistent processing conditions experienced by the wafers. The work in this thesis focuses on the measurement of this process variability using intrusive and nonintrusive sensors to measure the plasma parameters as accurately as possible. Statistical approaches are used to build correlations between etch rate variability and the sensor measurements. The main finding of the thesis concludes that the combination of appropriate process measurement with sensors and statistical algorithms provide a very powerful tool to a process engineer in diagnosing process variability.