Massachusetts Institute of Technology (MIT) has developed an adaptive, intelligent, monitoring system for large power transformers. Four large transformers on the Boston Edison system are under continuous surveillance by this system, which can summon attention to anomalous operation through paging devices. The monitoring system offers two advantages over more traditional (not adaptive) methods of tracking transformer operation: while temperature measurements have been made for some time and can even be incorporated into SCADA systems, comparison of actual measurements with predictions based on operating conditions and ambient temperature can provide a much more sensitive measure of actual transformer condition; and while oil samples are taken from transformers and analyzed for chemical species that might indicate abnormal operation, continuous monitoring of oil chemistry can give immediate indication of something going wrong, perhaps in time to remove the transformer from service and prevent a catastrophic failure. The authors describe the monitoring system and field experience.