Evidence for effects of PM2.5 on systemic oxidative stress in pregnant women is limited, especially in early pregnancy. To estimate the associations between ambient PM2.5 exposures and biomarkers of lipid peroxidation and total antioxidant capacity (T-AOC) in women with normal early pregnancy (NEP) and women with clinically recognized early pregnancy loss (CREPL), 206 early pregnant women who had measurements of serum malondialdehyde (MDA) and T-AOC were recruited from a larger case-control study in Tianjin, China from December 2017 to July 2018. Ambient PM2.5 concentrations of eight single-day lags exposure time windows before blood collection at the women’s residential addresses were estimated using temporally-adjusted land use regression models. Effects of PM2.5 exposures on percentage change in the biomarkers were estimated using multivariable linear regression models adjusted for month, temperature, relative humidity, gestational age and other covariates. Unconstrained distributed lag models were used to estimate net cumulative effects. Increased serum MDA and T-AOC were significantly associated with increases in PM2.5 at several lag exposure time windows in both groups. The net effects of each interquartile range increase in PM2.5 over the preceding 8 days on MDA were significantly higher (p < 0.001) in CREPL [52% (95% CI: 41%, 62%)] than NEP [22% (95% CI: 9%, 36%)] women. Net effects of each interquartile range increase in PM2.5 over the preceding 5 days on T-AOC were significantly lower (p = 0.010) in CREPL [14% (95% CI: 9%, 19%)] than NEP [24% (95% CI: 18%, 29%)] women. Exposure to ambient PM2.5 may induce systemic lipid peroxidation and antioxidant response in early pregnant women. More severe lipid peroxidation and insufficient antioxidant capacity associated with PM2.5 was found in CREPL women than NEP women. Future studies should focus on mechanisms of individual susceptibility and interventions to reduce PM2.5-related oxidative stress in the first trimester.