A warning system for drilling engineering has been developed to implement of real-time detection of work status and give early warnings about potential drilling accidents during well construction process. With the help of comprehensive logging instrument, this warning system can gather drilling engineering parameters (bit weight, torque, hook load el at). In this system those parameters can display on computer screen, which is convenient for engineers to observe parameters variation tendency. Mathematical statistics methods (average time, variance, amplitude, etc.) are taken into account to explain the reason of changing parameters curves. The most important principle of this warning system is to establish the relationship between drilling accidents and engineering parameters. The process of selecting abnormal drilling parameters is based on Grey Correlation Theory, which is an effective way to figure out some crucial abnormal parameters that cause a certain drilling accident. The application of this warning system in Shengli Oilfield of China shows that this system can exactly monitor abnormal drilling parameters and run smoothly with convenient operation. In addition, this system gives accurately early warning information of common drilling accidents (kick, well leakage, H2S abnormal, drilling tool failure etc.) either in oil field or in remote monitor center, and the warning accuracy is above eighty percent. Drilling accidents and complicated downhole conditions are always big threats during the well construction process. Failure to monitor abnormal drilling engineering parameters and failure to take effective measures to warning symptoms may have a significant impact on well economic benefit due to long non-productive time and high cost of accident treatment. Therefore it is much useful for this system to provide warning information of changing downhole conditions before an eventual incident occur, which helps decision makers to analyze downhole situation in advance and take actions immediately according to warnings.