Leveraging multi-core hardware can pose a challenge to any programmer, but especially when dealing with mature software. In and of itself, mature code can be challenging to interpret, with its strange constructs, GOTOs, and forgotten, quirky syntax that may even pre-date the ANSI C standard. The nature of this kind of code, written by mathematicians with serial execution in mind, can make the addition of OpenMP directives a daunting task for programmers. Accustomed to code that predictably executes the same instructions in the same order, programmers deal with the non-deterministic nature of parallel execution, along with error handling and thread safety. This presentation covers some of the decisions, trade-offs, and engineering effort required to parallelize code that had been untouched for years.