Based on Lyster and Ranta (1997), this current study made a comparison between native and non-native English speaking teachers' corrective feedback to students' errors, and also between different teachers' feedback types and students' uptake. The database consisted of 738 minutes' or 12.3 hours' classroom observation, including two types of teachers, six types of corrective feedback and two types of student uptake. Results showed that recast was the most frequently used feedback type across all the teachers, which simultaneously led to low rate of learner repairs; non-native English speaking teachers provided overwhelmingly more feedback than native teachers, and they tended to use more recast; both native and non-native teachers preferred to use varied kind of feedback at similar distribution which might suggest that corrective feedback did not necessarily rely on teacher types with students of the same proficiency level; elicitation tended to be the most effective feedback type in both native and non-native teachers' class, which might indicate that feedback types which can trigger negotiation of form were effective no matter what types of teachers use them. The results suggested that teachers should avoid using recast and opt for elicitation for more effective learning.