National human rights institutions (NHRIs) have proliferated over the last few decades. Today, they are regarded as part of a domestic infrastructure to facilitate the realization of human rights. Yet we know relatively little about whether NHRIs actually lead to the improvement of human rights. This thesis critically examines the NHRI phenomenon, by asking two questions: Do NHRIs improve states' respect for international human rights? And if yes, through which mechanisms? Theoretically, I draw a distinction between optimist and skeptic views of human rights institutions. The skeptical expectation is that NHRIs are adopted as window-dressing. Here, NHRIs are institutions of cheap talk with no means of power to back their rhetoric, and are therefore unable to bring about change. Within the optimist view, I spell out two possible mechanisms. A monitoring NHRI may improve human rights conditions in a state by making it more costly for the state to violate human rights. An NHRI working through socialization is part of a global process of norm diffusion. Here, states change their behavior not because of cost-benefit analyses, but because they adopt new norms of appropriate behavior. The empirical approach of this thesis consists of two parts. A time-series--cross-section statistical analysis maps global patterns of NHRI adoption, institutional characteristics, and human rights practices. A case study of the Norwegian NHRI from 1999 to 2015 picks up some central puzzles of the statistical analysis. Together, the analysis gives rise to three main conclusions. First, no evidence suggests that NHRIs have a global average positive impact on human rights practices. Second, where there nevertheless is a positive association between NHRI adoption and human rights, the patterns tentatively match the mechanism of socialization. Third, this effect seems to be strongest where a state has an A-accredited institution. The case of Norway shows how window-dressing can be a dominant mechanism, but where the concern with the NHRI's accreditation entered into a social dynamic of naming, shaming, and reputation---which ultimately led to to the creation of a new and independent NHRI in 2015. In summary, there is no place for unrestrained optimism towards the effectiveness of NHRIs. Adoption or accreditation of an NHRI may not lead to much alone. But this is not a call for cynical skepticism. On the contrary, it is a call for a careful, focused attention to what it actually takes to make NHRIs succeed in diverse contexts around the world.