abstract this study investigates the teachers’ correction of students’ spoken errors of linguistic forms in efl classes, aiming at (a) examining the relationship between the learners’ proficiency level and the provision of corrective feedback types, (b) exploring the extent to which teachers’ use of different corrective feedback types is related to the immediate types of context in which error correction takes place, and (c) investigating the relationship between corrective feedback and learner uptake. the data for this study were drawn from transcripts of 36 hours of naturalistic classroom interactions recorded from twenty four classes of three teachers (eight classes per teacher, four elementary and four pre-intermediate). a total of 1142 feedback moves were identified and coded based on lyster and ranta (1997) and farrokhi’s (2005, 2007) classifications to reflect corrective feedback types, uptakes, and context types. chi-square analysis was performed in order to analyze associations between variables of this study in pairs (that is, proficiency level and corrective feedback types, corrective feedback types and uptakes, corrective feedback types and context types). according to the findings, there were statistically significant associations between corrective feedback types and proficiency level and also between corrective feedback types and uptakes. it was found that marked recasts were the most frequent type of feedback in elementary level. also, negotiated feedback was the most favored feedback type provided by the teachers in pre-intermediate level. moreover, the most successful technique for ii eliciting uptake was found to be negotiated feedback in both proficiency levels. however, the findings indicated that type of context in which error correction took place did not affect type of corrective feedback used by l2 teachers for dealing with the learners’ non-target-like forms.