We reviewed and meta-analyzed 10 studies (N = 492) that examined the association between (risk for) child maltreatment perpetration and basal autonomic activity, and 10 studies (N = 471) that examined the association between (risk for) child maltreatment and autonomic stress reactivity. We hypothesized that maltreating parents/at-risk adults would show higher basal levels of heart rate (HR) and skin conductance (SC) and lower levels of HR variability (HRV) and would show greater HR and SC stress reactivity, but blunted HRV reactivity. A narrative review showed that evidence from significance testing within and across studies was mixed. The first set of meta-analyses revealed that (risk for) child maltreatment was associated with higher HR baseline activity (g = 0.24), a possible indication of allostatic load. The second set of meta-analyses yielded no differences in autonomic stress reactivity between maltreating/at-risk participants and nonmaltreating/low-risk comparison groups. Cumulative meta-analyses showed that positive effects for sympathetic stress reactivity as a risk factor for child maltreatment were found in a few early studies, whereas each subsequently aggregated study reduced the combined effect size to a null effect, an indication of the winner’s curse. Most studies were underpowered. Future directions for research are suggested.