This paper investigates post-obstruent final /r/ deletion in French (e.g., livre [liv(K)], votre [vot(K)]) using two large corpora and phoneme-level forced alignment to analyze both pronounced and deleted /r/ variants. The study examines factors influencing /r/ deletion, including phonetic context, speaker profession, gender, age, and speaking style. Results show that right-phone context, age, and speaking style significantly affect /r/ deletion, with specific patterns also observed for the preceding phone and speaker profession. Gender has no effect. With over 390 speakers producing 14,167 individual /r/ tokens, the study provides robust insights into phonetic and social factors shaping final post-obstruent /r/ deletion. The combination of speech technology, NLP methods, and linguistic analysis of large-scale naturalistic data improves our understanding of this phenomenon, challenging assumptions about gender’s relevance in this phenomenon.