Achievement goals (AGs) have been argued to influence general learning processes and outcomes, while their roles in L2 learning remain underexplored. This study, consisting of two sub-studies, aims to examine how different types of AGs (mastery-approach goals, performance-approach goals, mastery-avoidance goals, and performance-avoidance goals) contribute to overall and skill- (knowledge-)1 specific L2 achievement (curriculum-based course exam scores) and proficiency (Cambridge English proficiency test scores) using cross-sectional and longitudinal approaches. Participants of both sub-studies were 1,355 adolescent EFL learners from a junior high school in rural China. In Sub-study 1, the students participated in a questionnaire survey, an endterm English exam, and a Cambridge English proficiency test. Path analysis results showed that different AGs predicted overall and skill-specific L2 achievement and proficiency with variations. In Sub-study 2, the same participants completed three waves of questionnaire surveys and endterm exams spanned over three 6-month semesters. Cross-lagged panel modelling results indicated that AGs positively influenced subsequent achievement, but the effects were unstable; achievement positively and stably influenced subsequent AGs. Our findings provide implications for L2 teachers to help their students set goals to close their achievement gap and reach their full L2 potential.