Perhaps you've got lots of metadata at your institution -- some of it in relational databases, or collections of XML docs, or possibly even an assortment of spreadsheets. Maybe you've been hearing a lot about RDF, SPARQL, and the limitless utopian possibilities of Linked Data, and want to get in on the action. So how do you get your data out those stale old formats and serialized in RDF where it can run amok in the sunny green fields of the Semantic Web? This talk will break down the decisions you'll need to make and the pain points you'll encounter as you attempt to migrate your metadata from tables to triples, elements to predicates, and/or strings to URIs. We'll draw from our experiences as part of the Hydra Project's MODS and RDF Descriptive Metadata Subgroup, but this talk will be applicable to any schema or platform. Topics will include: choosing the right vocabulary and namespace for a particular data point; modeling complex XML hierarchies as a graph; how to understand what you can and can't do with a particular predicate's range; an overview of the status of various efforts to represent metadata schemas in RDF; dealing with the un-ordered-ness of RDF; and knowing when to compromise on data specificity so you don't drive yourself bonkers. Many of the above are rarely covered and documentation on them is fairly limited -- this talk is all about addressing topics that seem to have fallen through the cracks.