How does donor discourse on best and worst practices in foreign aid giving relate to their aid giving practice? While research has measured donor performance and ranked donors on the quality of their aid, we know little about the relationship between the international rhetoric on what makes aid giving effective and donors’ actual aid giving choices. In this project, we focus on a historically common donor practice that has been widely recognized as hurtful for aid effectiveness: that of “tying” foreign aid. Foreign aid is “tied” when a donor requires that portions of its aid be spent on goods and services from firms in its home country. We collect and clean a corpus of 161 OECD Development Assistance Committee (DAC) documents, including 40 annual Development Cooperation Reports (1980–2020) and 121 peer reviews (1996–2020). We employ quantitative text analysis methods to measure donor discourse on the topic of tied aid both at the year and at the donor-year level. We compare this discourse to data on prior and subsequent actual amounts of tied aid in aid commitments and disbursements over the past four decades. For all DAC donors, we assess whether the tied aid discourse in DAC documents (i) responds to the actual donor practice of aid tying and/or (ii) influences subsequent decisions of donors to tie their aid. We find robust evidence of the former and some evidence of the latter.