For hot rolled strip, a more flexible cooling control strategy is required to make the strip cooled according to the cooling path set in advance, and consequently take full advantage of the phase transformation and precipitation strengthening during the cooling process. Thus, a new method based on temperature observer for strip temperature control was proposed, in which strip temperature observer based on physic models was developed to monitor the strip temperature and, at the same time, the model parameters of the observer were revised according to the measured temperature. On this condition, an optimal controller was applied to optimize the valves setting in each cooling unit, as well as to ensure the minimum deviations between the observed temperature distribution curve and the desired one. Practical application shows that this method can not only control the strip temperature accurately, but also overcome the limitations of fixed cooling pattern and fulfill the objective of cooling path control.