The invention discloses a distributed remote vibration monitoring and fault diagnosis system of a large steam turbine-generator, comprising a database subsystem in which steam turbine-generator vibration data is stored, a workstation subsystem capable of providing real-time status monitoring, historical data analysis and fault symptom analysis, and a Web server subsystem provided with remote monitoring and fault diagnosing software, wherein, the subsystems are connected with one another by the Ethernet, and are integrated by a Web services technique. A method combining a traditional fault diagnosing technique based on Fourier analysis, a novel independent component analysis feature extraction technique, a Hilbert-yellow time frequency analysis technique, an empirical decomposition algorithm trend analysis technique and a self-organizing map neural network is adopted for monitoring and diagnosing in the system, thus improving the monitoring real time and the accuracy of diagnosis. The system has strong universality and reliability; and the system configuration can be adjusted, thus conveniently realizing distributed remote monitoring and diagnosis based on the internet as a medium.