Selective assembly is a method to achieve high precision fit using low precision parts. The current practices in selective assembly lack a physical process model to guide the design and process of mating parts, consequently, selective assembly is often a nightmare for manufacturing people. Process capability indices (PCIs) are the widely accepted process parameters, functioning as a measure of process performance. In this paper, the concept of PCI-based tolerance is proposed as an interface between quality requirements and statistical process control (SPC) parameters. An analytical model involving PCI-based tolerance is developed to predict and assure the matchable degree in selective assembly. As matchable degree and other process quality requirements can be assured at the stage of design by introducing PCI-based tolerance, the process of selective assembly can be improved significantly as an effective way to achieve precision assembly with economical manufacturing processes.