Documentation of knowledge about biological pathways is often informal and vague, making it difficult to efficiently synthesize the work of others into a holistic understanding of a system. Several researchers have proposed solving this problem by modeling pathways using formal computer languages, which have a precise and consistent semantics. While precise, many of these languages may be too low-level to feasibly model complex pathways. We have developed the Pathway Modeling Language (PML), a high-level language for modeling pathways. PML is based on a biological metaphor of molecules with binding sites and has special constructs for handling compartment changes in pathways. Our preliminary work has shown that PML's language constructs serve as a promising basis for modeling complex pathways in a readable and composable manner.