This article develops an efficient and accurate method for numerical evalua
tion of the integral equation which defines the American put option value f
unction. Numerical integration using Gaussian quadrature and function appro
ximation using Chebyshev polynomials are combined to evaluate recursive exp
ectations and produce an approximation of the option value function in two
dimensions, across stock prices and over time to maturity. A set of such so
lutions results in a multidimensional approximation that is extremely accur
ate and very quick to compute. The method is an effective alternative to fi
nite difference methods, the binomial model, and various analytic approxima
tions.