Second language (L2) learners can assess their oral skills with some accuracy. A question is whether learners can reliably use self-assessment to track their language gains over time. During the spring of two subsequent years, 80 university L2 Chinese learners completed an oral skills self-assessment based on the NCSSFL-ACTFL Can-Do Statements (ACTFL, [, 2015]) and took the computerized Oral Proficiency Interview (OPIc). Results were compared to see how accurate the self-assessments were and to determine if the students' self-assessment skills improved. Most students' OPIc gains were reflected in their self-assessment gains; meanwhile, learners tended to underassess rather than overassess their skills. Students at the Novice and Advanced proficiency levels were more accurate self-assessors than were Intermediate-level learners. There was no difference in self-assessment rating accuracy between Years 1 and 2. The findings suggest that self-assessment based on the Can-Do Statements is valuable for low-stakes assessments: for example, to monitor students' proficiency gains and to globally track the way in which language programs promote proficiency growth.