Krashen's (1981) second language learning monitor model and Labov's (1978) first language acquisition monitor model are compared, and it is concluded that monitoring is not an all-or-nothing phenomenon, but is variable. It is also suggested that the ability to monitor in formal language styles may improve accuracy in less formal styles, based on existing research. More research is recommended to discover whether nonnative speaker styles stratify as clearly as do Labov's native speaker styles. It is suggested that Labov's monitor model is a more accurate representation of the process than Krashen's, giving these pedagogical implications: learning activities have a legitimate place in the foreign language syllabus; and the social situation of speaking can affect students' accuracy, causing use and even overuse of the monitor in class. It is noted that despite the superiority of Labov's model in this respect, Krashen's acquisition-learning distinction, and the resulting mix of activities, is important in the language classroom. (MSE)