Previous chapters have discussed validating typed data in data frames, database tables, and more complex data structures. Serialized data, such as that in CSV and other flat files, typically has no type information and the values themselves can be ambiguous. This complicates matters. This chapter discusses common variations among flat-file formats and describes metadata files that can be used to reduce or eliminate ambiguity. It introduces the tdda.serial tools and functions for working with its own metadata files, as well as with CSVW and Frictionless metadata files. The chapter closes with a discussion of non-tabular data and of using Pydantic to validate such data.