This paper presents a primer on building a scientifically oriented teacher evaluation instrument. It stresses the importance of accurate measures and accepts the presupposition that scientific approaches provide the most accurate measures of student teacher performance. The paper discusses the scientific concepts of validity and reliability, and provides a checklist for assessing student teacher observation instruments. The checklist suggests two questions relating to validity: (1) Is there ample evidence that the items chosen are related to pupil learning (content validity); and (2) Are the items used in the instrument related to any selected theory of teaching and learning (construct validity)? In reference to the reliability of the instrument, seven criteria are suggested: adequate training of those using the instrument; items stated in overtly behavioral terms; items stated positively; items written in the singular; only one type of behavior per item; items written in present tense; and all items scored using a 5-category forced choice procedure. (Contains 14 references.) (NAV)