According　to　the　shortage　of　conventional　PID　in　thermal　cascade　control,　this　paper　adopts　off-line　and　online　Immune　Genetic　Algorithm　(IGA)　to　optimize　PID　controller,　respectively.　Experimental　results　show　that　off-line　IGA-PID　reduces　overshoot　effectively,　and　online　IGA-PID　not　only　eliminates　overshoot,　but　also　shortens　settling　time　sharply.　These　experiments　also　prove　that　online　immune　genetic　PID　has　a　better　application　in　optimizing　thermal　process　control　which　has　large　inertia　and　nonlinear　systems.　©　2014　Taylor　&　Francis　Group,　London.