Article gives methodical material for e-service architecture selection as multicriterial optimization task. Initially problem position is described – e-service architecture as Web service graph selection of multiple available solutions. To solve this problem, algorithm graph is defined as a mathematical model for transformations. Reduction from the algorithm graph to the Web service graph is described. Five Web service graph estimation criterions are defined (Costs of Production, Time of Execution, Exploitation Costs, Reusability and Integrity). Specific architecture solution is multicriterial optimization task that is reduced to detecting the Pareto set. Selected E-service architecture is a Web service graph that belongs to detected Pareto set. The given approach is demonstrated on example base and results are displayed in article.