This paper examines how domestic and international financing for HIV is, and ought to be, distributed. We build a theoretical framework that decomposes domestic and international financing for HIV into nonlinear functions of national income, HIV prevalence, and government effectiveness. We test this model, paying particular attention to nonlinearities and to problems of bad controls, multicollinearity, and reverse causality. Finally, we use the fitted values of quartile regressions to study how much countries could reasonably pay domestically and how much they should receive from donors. Worryingly, countries with higher financial means receive on average more aid per PLHIV than very poor ones, and countries with higher HIV prevalence receive on average less aid per people living with HIV. The normative analysis concludes that US$3.08 billion of fiscal space could be created in LIC and MIC. We identify the countries that could be allocated more aid.