Monitoring of process operations, quality control, and fault diagnosis are important components of the PAT initiative. Multivariate statistical methods provide techniques that detect abnormalities in process operation. Process information can then be interpreted to diagnose the source cause of the abnormality. We have developed a process monitoring and diagnosis system, integrating statistical and knowledge-based techniques for multivariable batch processes. Its deployment is discussed via case studies with simulated batch penicillin fermentation.