The paper describes an algorithm for computing the least weighted squares estimator, discusses computational aspects and modifies the Durbin-Watson test for the context of the least weighted squares.