Pronunciation is a vital component of oral language proficiency, yet it remains underrepresented in English language instruction, particularly within higher education EFL/ESL contexts. This study aims to address this pedagogical gap by developing and validating a non-test self-assessment instrument to measure pronunciation self-efficacy among intermediate English language learners (ELLs). Grounded in constructivist and learner-centered theories, the instrument integrates four dimensions: Phonological Awareness (PA), Accent and Intonation Awareness (AIA), Self-Confidence (SC), and Self-Monitoring (SM). A total of 43 university students participated, selected via purposive sampling. The instrument, comprising 20 Likert-scale items, underwent content validation by three experts, followed by psychometric evaluation through Rasch analysis using Winsteps software. The results demonstrate strong psychometric properties across all dimensions. Item and person reliability values ranged from moderate to high (0.64–0.90), with Cronbach’s Alpha (KR-20) between 0.68 and 0.90, indicating internal consistency. Person separation indices varied from 1.33 to 2.70, suggesting sufficient ability discrimination. Item fit statistics were within acceptable thresholds, supporting construct validity. Principal Components Analysis (PCA) of residuals confirmed unidimensionality, particularly for the SC construct, where Rasch explained variance reached 70.6%. PT-measure correlations further supported item discrimination quality. The validated instrument fills a methodological gap in the assessment of pronunciation self-efficacy and offers a practical, learner-centered tool for pedagogical implementation. It contributes to the advancement of self-regulated learning in pronunciation and opens pathways for further research into affective-cognitive correlates of pronunciation in EFL contexts.