The study compared the effects of two types of form-focused written feedback--direct corrective feedback (DCF) and metalinguistic explanation (ME) given to the whole class--on Japanese university students' accuracy of use of two grammatical structures: indefinite article and the hypothetical conditional. Both types of feedback were given with and without an opportunity to rewrite. Accuracy of use was measured in new pieces of writing. The feedback led to increased accuracy for the hypothetical conditional but not for the indefinite article. The effectiveness of the DCF proved longer lasting than the ME. Also, providing opportunity for revision enhanced the effect of the feedback. Overall, DCF followed by revision proved the most effective type of feedback. The results suggest that when form-focused written feedback is directed at two features that vary in saliency and complexity, learners are likely to focus on the structure that contributes more to the global meaning of the text. The results also indicate that directly correcting the errors learners make with respect to a complex syntactical structure is more beneficial than giving them a metalinguistic explanation.