The objectives of this study are twofold: (1) uncovering the local criteria for evaluating language-teaching performance, and (2) unraveling and conceptualizing language teachers' perception of these criteria. To this end, the study used grounded theory to collect and analyze interview data from twelve experienced language teachers who were willing to share their perception of evaluation criteria with the researcher. Instead of starting with a pre-specified statistical sample of subjects, the study started with a general question and sampled theoretically relevant concepts, and stopped data collection when the core category and its related concepts and categories were saturated. Since the concerns of the teachers were a recurrent theme, in reporting the results each criterion was juxtaposed with language teachers' concerns related to that criterion. The results can help supervisors make more informed decisions concerning the evaluation of language-teaching performance, and add a series of data-driven, context-sensitive propositions to the knowledge base of language-teacher evaluation.