The invention discloses a self-adaptive random test method based on iteration area equipartition and positioning, which is used for reducing blind generation of test cases in a random test and improving the efficiency of the random test. The method mainly comprises the steps that 1, determining the range of an input domain; 2, randomly generating a first test case in an input domain; 3, equally dividing the input domain, randomly selecting a sub-domain at the same time, then equally dividing (hypothetical equally dividing) again, randomly generating test cases for the sub-domains which are equally divided again, and forming a candidate test case set; and 4, aiming at the candidate test case set, determining an adjacent region set, forming an executed test case set by the test cases in theadjacent region, and determining the next test case to be executed by adopting an FSCS _ ART (self-adaptive random algorithm of a candidate set with a fixed size). The process is continuously repeateduntil program errors are found. Experimental verification proves that compared with a random test method, the method has the advantage that the performance is improved by 30%.