In monitoring complex systems, apart from quick detection of abnormal changes of system performance and key parameters, accurate fault diagnosis of responsible factors has become increasingly critical in a variety of applications that involve rich process data. Conventional statistical process control (SPC) methods, such as interpretation and decomposition of Hotelling’s T2-type statistic, are often computationally expensive in such high-dimensional problems. In this article, we frame fault isolation as a two-sample variable selection problem to provide a unified diagnosis framework based on Bayesian information criterion (BIC). We propose a practical LASSO-based diagnostic procedure which combines BIC with the popular adaptive LASSO variable selection method. Given the oracle property of LASSO and its algorithm, the diagnostic result can be obtained easily and quickly with a similar computational effort as least squares regression. More importantly, the proposed method does not require making any extra t...