This assumption is especially characteristic of neoclassical economists. Some non-neoclassical economists do also accept it, but some do not.
Neoclassical economists usually assume, in other words, that human beings make the choices that give them the best possible advantage, given the circumstances they face. Circumstances include the prices of resources, goods and services, limited income, limited technology for transforming resources into goods and services, and taxes, regulations, and similar objective limitations on the choices they may make.
I should modify that a little right away. Strictly speaking, neoclassical economics does not assume that real, concrete human beings are rational and self-interested. Rather, most economists assume that economic systems work as if they consisted of rational, self-interested persons. After all, it is averages that count for these purposes. People are of all sorts; sneaky and altruistic, smart and dumb, but if the average is a person who is rational and self-interested, then the system will "act as if" people in general were rational and self-interested. At least, that is the basis of "neoclassical" economics: it is assumed that deviations from rational self-interest are random and will cancel out and so the system will act as if everyone were rational and self-interested. Accordingly, neoclassical economics studies an economic system consisting of rational, self-interested persons.
We should pause a little further over these assumptions before moving on. They are not particularly common-sense assumptions. We all know of examples of non-self-interested behavior -- of people who give to the church and to good causes and who sacrifice themselves in other ways -- and my common sense (at least) suggests to me that people are often irrational chumps.