Drilled well control is a process that incorporates the assessment of well status through the monitoring of its physical parameters. This management allows the detection of well anomalies such as gas kick and mud loss. In this paper is developed a gas kick/ well loss early detection model that determines the well condition and early predicts possible anomaly. The developed model insures the system safety along the drilling operation through the early prediction of gas kick influx or mud loss using a model-based control system solution. A reduced order model is derived to predict the mud pressure and flow rate responses for given real time pre-filtered measurements. The model sensitivity to the well anomalies is captured through its coefficients variations permitting a real time evaluation of well status by monitoring the model coefficients location. The obtained results prove that the presented solution is capable of detecting a gas kick before the gas influx reaches the surface in contrast with the widely used delta flow method capturing the kick once the gas is on the floor level.