Cost-effective software project management has the serious need to focus resources on those areas with highest criticality. Identifying such components in advance that need high development effort or that are likely to produce many failures during operation and assigning additional design or corrective effort is one approach for effective resource allocation. Complexity metrics are applied during the development of large telecommunication software in order to identify high risk components and to tailor reliability growth models. Five classification techniques (Pareto classification, classification trees, factor-based discriminant analysis, fuzzy classification, neural networks) are compared for identifying critical components. For testing and maintenance phases we combined this approach with tailored reliability growth models. Results from a current large-scale switching project are included to show practical benefits.