The high throughput and operating cost of the reactor provide the economic incentive in considering the on-line optimization of the LDPE tubular reactor. For an on-line optimization procedure it is crucial to develop a flexible and adaptive on-line parameter estimator, which based on the appropriate measurements from the reactor will adjust certain key model parameters so that the model can capture the actual reactor operation and reliably be used for the real-time optimization of the reactor performance. Various issues involved in developing the on-line estimator for such applications as the mathematical modeling of the reactor, the selection of the adjustable model parameters and reactor measurements, the decomposition of the problem according to the characteristic time constants of various disturbances, and the efficient and robust solution of the mathematical problem are discussed. The tools developed were successfully used on an EXXON high-pressure LDPE tubular reactor. Results from the numerical experiments that are compared with scaled data from a real reactor show that the estimated temperature profiles agree remarkably with those observed at various phases of the reactor operation.