The present study investigates whether the type of feedback (direct or indirect) given to 41 intermediate EFL learners on six types of errors (verb tense, noun ending, word choice, sentence structure, article and preposition) results in improved accuracy both in the short and long terms. To this purpose, English students in a writing class at the University of Guilan divided randomly into direct, indirect and control groups were presented with the same topics to write about. The errors on pretest, posttest 1 and posttest 2 in each group were detected and the means of errors were investigated by ANOVA analysis. At last, the study found a significant effect for the indirect feedback on accuracy improvement in the use of the verb tense, noun endings, sentence structure and articles in both short and long term periods, but no significant effect for the preposition in the long term and word choice in either time periods. Although the direct feedback significantly affected the accuracy level for the use of noun endings, prepositions, articles and sentence structure in the short term, it did not demonstrate any statistical significance in the long term. Moreover, no effect on verb errors and wrong words were found in either time periods. Finally, the indirect group outperformed the direct one on accuracy improvement for the total number of errors. The study also considered the use of avoidance strategy due to the provided corrective feedback and found that, in spite of the probability of the tendency towards using it, providing corrective feedback is still necessary in improving L2 writing accuracy.