A study was undertaken as part of the process of trying to develop an instrument for systematically coding reading and other language arts activities in the secondary content area classroom. Six teachers chosen from five secondary schools were each observed while teaching four randomly selected classes over a two-week period. All observations were audiotaped and nonverbal activity was recorded in a log for use later by coders. Of the 22 categories in the schedule, 13 proved to be too narrow or too uncommon to discriminate among the teachers. In three other categories, behaviors occurred frequently but were so similar as to fail to distinguish among teachers. Six behaviors occurred with enough frequency and variety among teachers to justify calculating generalizability (reliability) coefficients for the categories. Data from this study suggest that reading and listening activities do occur in the content classroom with enough frequency and variety that reading and listening activity categories can be devised that will be generalizable. (TJ)