This article presents a solution to the hydrothermal scheduling (HS) problem by evolutionary programming (EP). The purpose of HS is to minimize fuel costs of thermal units by allocating hydro and thermal units optimally, satisfying constraints on system operation. EP performs better than conventional methods in converging to near optimum results in the solution of the HS problem, which consists of nonlinear objective function and constraints. In EP, offsprings are generated from randomly generated initial parent vectors by Gauss or Cauchy mutations. Parent vectors and their offspring vectors compete with each other. Better individuals are selected as new parent vectors for the next iteration. As the iteration makes progress, convergence to optimum solution increases. In this study, instead of stochastic competition in existing EP algorithms, deterministic competition is utilized. The control parameter "scaling factor" is taken as variable instead of constant. Thus, better results have been obtained in convergence rate, solution time and success rate.