Purpose: Phonetic training has been found to be an effective way on second language (L2) learning, but the evidence is not conclusive regarding the effectiveness of different instructional approaches and possible interactions between language learners, training features, and outcome measures. This study aims to meta-analyze existing studies to provide an accurate estimate of the overall effect and investigate factors that may moderate its effectiveness. Method: We conducted a systematic search in major databases, identified seven potential moderators, and conducted a random-effects model meta-analysis for each variable. Results: A comprehensive literature search yielded 65 primary studies that involved 2,793 L2 learners, generating 223 effect sizes for between-groups contrasts. Results showed that phonetic training had a large positive effect on the enhancement of L2 phonetic competence, d = 0.762. Subsequent moderator analyses revealed that perceptual training showed a larger mean effect size compared to production training and combined training; phonetic training at the high school level exhibited the largest mean effect size among all educational levels; phonetic training, either perceptual training or production training, had a greater impact on improving learners' perception competence than production competence; and outcome measured by identification tasks generated the largest effect, followed by the combination of discrimination and identification tasks, subjective perception judgment, and discrimination tasks, while objective acoustic measurement yielded the smallest effect size. Conclusions: Our study provided a quantitative synthesis of studies investigating the efficacy of L2 phonetic training and examined various moderating variables, which indicated the heterogeneity and limitations of research on this topic. The results highlighted the need for further investigation of the potential factors of L2 phonetic training and the relationship between perception and production.