While some legal rules are formulated precisely enough to be implemented as programs, much of the law today is far from being computable. Given the similarities between law and software, however, even if legal rules are not currently computational in nature, we can leverage principles from computer science to assess their quality and pinpoint potential improvements. Therefore, adapting the idea of code smells from the software engineering subfield of refactoring, we initiate the study of law smells, i.e., patterns in legal texts that pose threats to the comprehensibility and maintainability of the law. With five intuitive law smells as running examples – namely, duplicated phrase, long element, large reference tree, ambiguous syntax, and natural language obsession –, we develop a comprehensive law smell taxonomy. This taxonomy classifies law smells by when they can be detected (static vs. dynamic), which aspects of law they relate to (e.g., structure or grammar), and how they can be discovered (local vs. non-local). We introduce text-based and graph-based methods to identify instances of law smells, confirming their utility in practice using the United States Code as a test case. Our work demonstrates how ideas from software engineering can be leveraged to assess and improve the quality of legal code, thus drawing attention to an underappreciated area in the intersection of law and computer science.