One of the central problems in software engineering is its inherited complexity. It is recognized that cognitive informatics plays an important role in understanding the fundamental characteristics of software. This paper models the cognitive weights of basic control structures of software, and develops a new concept of cognitive functional size for measuring software complexity. Comparative case studies between the cognitive functional sizes and physical sizes of 20 programs are conducted. It is found that for a given design, although the physical sizes of software may vary in a wide range, the cognitive functional sizes are much stable and independent from implementation technologies and programming languages. Therefore, the cognitive functional size provides a novel, stable, and practical software complexity measurement and is helpful to explain the fundamental nature of software complexity in the phase of design, implementation, or maintenance in software engineering.