ABSTRACT Numerous methods of evaluating language teaching and learning are in common use. One accurate method is direct classroom observation, the most widely used form of which is interaction analysis. Since interaction analysis and FLint have been found wanting (lack of reliability, observer bias, multiplicity of categories, cumbersome recording system), the time-interval recording system (a method employed in social science studies) is offered as a practical and reliable alternative. Features of this system include: describing behaviors to be observed, recording occurrences of behavior in a time interval (usually ten seconds), drawing up an observational code, assessing accuracy by deriving inter-observer reliabilities, plotting data collected on a progress chart, and interpreting findings. Teachers, supervisors, and researchers may then effect changes in teaching procedures and programs based on their analysis of classroom observations.