This chapter describes commonly used methods for PCG, both methods primarily used for runtime generation in roguelikes and other PCG-heavy games, and methods that are currently mostly explored in research labs. The chapter is organized according to the type of method. Constructive methods, which are diverse, content-specific, and generally fast, are presented first. Search-based methods, presented next, represent the application of evolutionary computation methods such as generic algorithms and quality diversity to PCG problems. The chapter ends with a thorough coverage of newer PCG methods built on machine learning, including supervised, self-supervised, and reinforcement learning.