The prevalence of frailty varies among socioeconomic groups. However, longitudinal data for the association between subjective social status and frailty is limited. In this study, we examined whether subjective social status was associated with incident frailty. Data were obtained from a 14-year cohort of Chinese men and women (N = 694) aged 65 years and older who participated in the MrOs study—a longitudinal study on osteoporosis and general health in Hong Kong. Subjective social status at baseline (2001–2003) was assessed using a 10-rung self-anchoring scale. Incident frailty at the 14-year follow-up (2015–2017) was defined as proposed by Fried and colleagues. Ordinal logistic regressions were used to examine the association between subjective social status (high, middle, or low) and incident frailty. After adjustment for age, sex, marital status, objective socioeconomic status, medical history, lifestyle, mental health, and cognitive function, subjective social status at baseline was negatively associated with risk of developing frailty over time (OR 2.3, 95% CI 1.2–4.6). In sex-stratified analysis, the social gradient in frailty was only found in men. Social inequality in frailty in men but not in women supports interventions specific to gender inequality and frailty.