Great thermal stresses and fatigue damages will be developed during the startup of steam turbines, which will threaten the safety of operation. To save energy and improve the flexibility of the power unit, inlet steam temperature of steam turbines should be controlled online in an optimal way. A new method to obtain online optimal control schemes of the inlet steam temperature considering low cycle fatigue is presented in this paper using the Green's Function Method and the Pontryagin's Maximum Principle. New analytical models of temperature and thermal stresses are proposed. For a hot startup, constant material properties are used and the steam temperature history that can maintain maximum Von Mises stress close to the permitted value is proved to be the optimal control scheme. For a cold or warm startup, the optimization thermal stress considering temperature dependent material properties is found to be determinated by material properties and Green's functions, which may not be equal to the maximum permitted value. Application of the proposed method to the cold start-up process of a 600 MW steam turbine is introduced. Compared with the conventional start-up scheduling, the proposed optimal control scheme can shorten the time of startup greatly without exceeding permitted fatigue damage. (C) 2016 Elsevier Ltd. All rights reserved.