The computational properties of an econometric method are fundamental determinants of its importance and practical usefulness, in conjunction with the method's statistical properties. Computational methods in econometrics are advanced through successfully combining ideas and methods in econometric theory, computer science, numerical analysis, and applied mathematics. The leading classes of computational methods particularly useful for econometrics are matrix computation, numerical optimization, sorting, numerical approximation and integration, and computer simulation. A computational approach that holds considerable promise for econometrics is parallel computation, either on a single computer with multiple processors, or on separate computers networked in an intranet or over the internet.