The current study examined how form-focused instruction (FFI) with and without corrective feedback (CF) as output enhancement facilitated second language (L2) perception of Mandarin tones at both the phonetic and phonological levels by 41 Cantonese learners of Mandarin. Two experimental groups, FFI only and FFI-CF, received a 90-min FFI treatment designed to encourage them to notice and practice the categorical distinctions of Mandarin tones through a range of communicative input and output activities. During these activities, the instructors provided CF only to students in the FFI-CF group by recasting and pushing them to repair their mispronunciations of the target features (i.e., output enhancement). The control group received comparable meaning-oriented instruction without any FFI. The effectiveness of FFI was assessed via a forced-choice identification task with both trained and untrained items for a variety of tonal contrasts in Mandarin (high-level Tone 1 vs. mid-rising Tone 2 vs. high-falling Tone 4). According to statistical comparisons, the FFI-only group attained significant improvement in all lexical and tonal contexts, and such effectiveness was evident particularly in the acquisition of Tone 1 and Tone 4--supposedly the most difficult instances due to their identical phonological status in the learners' first language, Cantonese. The FFI-CF group, however, demonstrated marginally significant gains only under the trained lexical conditions. The results suggest that FFI promotes learners' attentional shift from vocabulary to sound learning (generalizable gains in trained and untrained items) and facilitates their access to new phonetic and phonological categories. Yet the relative advantage of adding CF to FFI as output enhancement remains unclear, especially with respect to the less experienced L2 learners in the current study.