Altered stress responsiveness is a risk factor for mental and physical illness. In non-pregnant populations, it is well-known that anxiety can alter the physiological regulation of stress reactivity. Characterization of corresponding risks for pregnant women and their offspring requires greater understanding of how stress reactivity and recovery are influenced by pregnancy and women’s anxiety feelings. In the current study, women were presented repeatedly with mental arithmetic stress tasks in the first and third pregnancy trimester and reported their trait anxiety using the state trait anxiety inventory. Cardiovascular stress reactivity in late pregnancy was lower than reactivity in the first pregnancy trimester (heart rate (HR): t(197) = 4.98, p < .001; high frequency heart rate variability (HF HRV): t(196) = −2.09, p = .04). Less attenuation of stress reactivity occurred in more anxious women (HR: b = 0.15, SE = 0.06, p = .008; HF HRV: b = −10.97, SE = 4.79, p = .02). The study design did not allow the influence of habituation to repeated stress task exposure to be assessed separately from the influence of pregnancy progression. Although this is a limitation, the clear differences between anxious and non-anxious pregnant women are important, regardless of the extent to which differing habituation between the groups is responsible. Less dampened stress reactivity through pregnancy may pose long-term risks for anxious women and their offspring. Follow-up studies are required to determine these risks.