This study examines the realization of Mandarin neutral tone (NT) in spontaneous conversational speech. Drawing on 8,550 phrase-medial NT tokens surrounded by citation tones, we investigate whether robust coarticulatory influences from adjacent tones persist in natural discourse and compare two subtypes of NT—one with a nasal offset and one without—thereby testing the combined effects of syllable structure and grammatical function. Our analysis confirms that adjacent tones exert a dominant influence on NT realization: ultrashort NT syllables largely continue the preceding tone’s contour, while normal- to long-duration syllables exhibit clearer transitions toward the subsequent tone. NT with a nasal offset maintains a higher pitch and greater F0 dynamics while remaining stable across durations, likely link to its stronger prosodic and grammatical cues. These findings highlight the interplay of tonal coarticulation, duration, and structural patterns in shaping NT variation.