Despite the existence of about one billion illiterates in the world, adult literacy programs make up 1-5 percent of government or donor budgets, and they remain severely underfunded in comparison to primary education. Though dropout and course completion rates improved in the 1990s, the outcomes of literacy instruction are still modest and may have improved little since the 1970s. The results may disappoint governments and donors who expect that once taught, people will have usable skills and remain literate. The modest results make it hard to increase coverage and to argue for increased expenditures for this sector. The results may be due to inefficient instruction but also to the structure of human memory, which has important implications for adult literacy acquisition. The need to learn the rapid recognition of complex patterns poses problems that are not apparent to people who became expert readers in their childhood.