What Is Dead Code?

Dead code is one of the few forms of technical debt that can be safely eliminated with zero functional impact — by definition, removing dead code changes nothing about what the software does. Despite this, dead code accumulates in every long-lived codebase because removing it requires knowing it is dead, and that knowledge is difficult to obtain without systematic analysis.

The cost of dead code is indirect but real. Developers waste time reading, understanding, and maintaining code that does nothing. Search results return dead code alongside live code, creating confusion. Refactoring efforts must account for dead code paths. Build times increase as dead files are compiled. Test coverage metrics are diluted by dead code that is either untested (lowering coverage) or needlessly tested (wasting test execution time).

In large codebases, dead code can represent 10-20% of total lines of code — a significant maintenance burden that can be eliminated safely once identified.

Dead code detection works by constructing a complete dependency graph and then identifying nodes (files, functions, classes) with zero in-degree — meaning nothing in the active codebase references them.

File-level dead code detection identifies files that are never imported by any other file and are not entry points (main files, scripts, test files, or configuration files that are invoked externally).

Function-level dead code detection identifies functions that are defined but never called within the reachable code graph. This requires call graph analysis in addition to import graph analysis.

The analysis must account for legitimate zero-in-degree files: entry points, CLI scripts, test files, migration files, and configuration files that are invoked by external systems. These are excluded from dead code findings through entry point detection heuristics and user-configurable exclusion patterns.