Abstract Traditional statistical process control (SPC) mainly focus on multivariate quality characteristics. With rapid advances in sensing technology and data acquisition systems, data in the manufacturing industry can be high-dimensional. It brings new challenges in process monitoring due to the well-known "curse of dimensionality". Thus, this thesis proposes several statistical methods to monitor the location and the covariance matrix of the high-dimentional processes. This thesis first focues on the monitoring of high-dimentional process mean. Minkowski distance-based charting statistic is developed to monitor individual observation when only few historical data are available. By using the dynamic control limits, the chart is shown to have geometric distributed run length. To overcome the problem of limited historical observations, self-starting procedure is incorporated to handle sequential monitoring by simultaneously updating the parameters estimation and checking for out-of-control conditions. Simulation results show that the proposed control chart is powerful for various kinds of shifts. It is also effective under heavytailed and skewed distributions. When observations are from a subgroup each time, the inter-point distance (IPD) is employed to measure the deviances of process mean from traget. The inter-point distance can make full of diatance information. Also, it is robust to outliers. The IPD-based chart represents a family of control charts with different values of d. Moreover, I have proved that the IPD-based chart is distribution-free under symmetric distributions. Then, we analysis the process variability by monitoring the changes in covariance matrices. An EGEN chart based on eigenvalues has been proposed, which works for both individual and subgroup observation. The EGEN chart is effective when dimension is larger than the number of observations, but the underlying distribution is limited to multivariate normal distribution. Therefore, by composing a L0 constraint on the optimization procedure, a spare Leading-Eigenvalue-Driven control chart is suggested. The control chart can acheive high detection powers for nomal and non-normal underlying distributions. Keywords: Distribution-free control charts, High-dimensional, Eigenvalue, Minkowski distance, Inter-point distance