Two experiments collected third-party evaluations from U.S. 4–5-year-old children (N = 80) who heard stories about caregivers helping or hindering their infants’ achievement of safe, dangerous, or ambiguous goals. Children’s evaluations were sensitive to danger: They switched from positively evaluating parents who helped access safe objects, to negatively evaluating those who helped access dangerous objects. Older children offered robustly positive evaluations of parents who protectively hindered access to dangerous objects, but younger participants were more likely to negatively evaluate these parents. Given a moderately risky goal that participants themselves judged as unsafe, children’s evaluations of helping and hindering were mixed, though there was preliminary evidence of a developmental shift. These findings show that young children go beyond basic inferences about whether an act promotes or hampers another agent’s goal when considering whether the action was good or bad. Instead, young children consider the broader consequences for the target’s welfare.