This study aimed to study the types of recasts given by the teachers and the rate of learners' uptake during one-on-one online English language teaching sessions in Thailand. It examined whether the recasts given to learners with different language proficiency level are different and whether those groups of learners respond to recasts differently. Participants were 30 Thai native adult learners of English at an online English school which were divided into three levels: beginning learners, intermediate learners, and advanced learners. Recorded teaching sessions in total of 30 hours were transcribed and analysed using One-way ANOVA, T-test, Chi-square, and Fisher's exact test. The results showed that teachers were likely to give more recasts to less advanced learners; however, most recasts given to all learners were similar in characteristics. Most recasts were word-length, declarative, grammar-related, substitution, unstressed, one change, and with no repetition. It was also found that more advanced learners could take advantages of being corrected more as the rate of their uptake was very high. Beginning learners were likely to respond to only some features of recast such as word-length, declarative, and substitution. This study could be beneficial for teachers to be aware of the suitable types of recasts which encourage learners with different language proficiency levels to repair their utterances after receiving recast moves.