Longitudinal evidence of long-term ozone exposure on heart rate variability (HRV, an early indicator of cardiovascular damage) is lacking and the potential mechanism remains largely unclear. Our objectives were to evaluate the cross-sectional and longitudinal associations of ozone exposure with HRV alteration, and the potential roles of protein carbonyl (PC, biomarker of oxidative protein damage) and transforming growth factor (TGF)-β1 in this association. This repeated-measures prospective study included 4138 participants with 6617 observations from the Wuhan-Zhuhai cohort. Ozone concentrations were estimated using a high temporospatial resolution model for each participant. HRV indices, PC, and TGF-β1 were also repeatedly measured. Cross-sectional and longitudinal relationships of ozone exposure with HRV alteration were evaluated by linear mixed model. Cross-sectionally, the strongest lag effect of each 10 ppb increment in short-term ozone exposure showed a 12.40 %, 8.47 %, 4.31 %, 8.03 %, 3.69 %, and 2.41 % decrement on very low frequency (VLF, lag 3 weeks), LF (lag 2 weeks), high frequency (HF, lag 0–7 days), total power (TP, lag 2 weeks), standard deviation of all normal-to-normal intervals (SDNN, lag 3 weeks), and square root of the mean squared difference between adjacent normal-to-normal intervals (lag 2 weeks), respectively. Longitudinally, each 10 ppb increment of annual average ozone was related with an annual change rate of −0.024 ms2/year in VLF, −0.009 ms2/year in LF, −0.013 ms2/year in HF, −0.014 ms2/year in TP, and −0.004 ms/year in SDNN. Mediation analyses indicated that PC mediated 20.77 % and 12.18 % of ozone-associated VLF and TP decline, respectively; TGF-β1 mediated 16.87 % and 27.78 % of ozone-associated VLF and SDNN reduction, respectively. Our study demonstrated that ozone exposure was cross-sectionally and longitudinally related with HRV decline in general Chinese urban adults, and oxidative protein damage and increased TGF-β1 partly mediated ozone exposure-related HRV reduction.