To date, conflict exists in the literature on whether or not and how teachers should react to EFL learners' written grammar errors. Coded versus uncoded corrective feedback has been one of the rarely explored areas of investigation in SLA. To shed light on the factors that may explain such conflicting results, this study investigated the effect of coded and un-coded written corrective feedback with regard to possible improvements in the accuracy in writing of preintermediate EFL learners. It, further, sought whether such an effect would last in the long run. In the course of 14 weeks, learners' errors in 2 groups (i.e., coded and uncoded) were reacted. A paired-samples t-test was run to analyze the obtained data. Analysis of the written pieces in the immediate post-test and delayed post test revealed that coded corrective feedback, compared to the uncoded group, had a significantly more positive influence on learners' accuracy improvement both in the short and in the long run. The findings imply that teachers should weigh the learners' abilities and interlanguage, proficiency level, and type of error before applying different feedback types. Moreover, the implications are discussed in terms of effective guidelines for teaching writing in EFL contexts.