This paper combines good frames and tameness and shows that a good \lambda-frame plus \lambda-tameness for 1- and 2-types implies the existence of a good \geq \lambda-frame. A detailed proof is given of Shelah’s result that a good \lambda-frame implies the uniqueness of limit models in \lambda.
It has since been shown that the use of tameness for 2-types is unnecessary by myself and Vasey; and Jarden (independently). See Boney and Vasey, Tameness and Frames Revisited for more details.