Guidelines and Standards for Setting Code Coverage Targets in Software Testing

For most projects with coverage metrics, a reasonable system‑testing goal is 70‑80% code coverage, while higher‑risk or safety‑critical systems may require 80‑90% or more, especially for unit tests.

Target selection should consider:

Failure cost : Increase targets for safety‑critical or high‑cost‑failure systems (e.g., medical, automotive).

Resources : Reduce targets if testing staff are insufficient or lack training.

Testable design : Raise targets when the system allows direct access to internal functions.

Development‑cycle state : Lower targets for legacy systems without original designers.

Many projects do not set a strict minimum coverage; they use coverage analysis mainly to identify untested areas and avoid over‑estimating coverage early on.

Achieving 100% coverage is often unrealistic and uneconomical; high coverage does not guarantee fault detection, and the cost of exercising rarely‑triggered test cases can outweigh the benefits.

Key reasons full coverage is impractical include high cost of low‑probability test cases, hidden errors in lower‑level code, future‑code accessibility, and shared code not exercised by all projects.

When coverage goals become the primary focus, the motivation shifts away from fault detection.

Coverage metrics commonly used for control‑flow analysis are:

Statement (line) coverage

Basic‑block coverage

Decision (branch) coverage

Condition/decision coverage

Modified condition/decision coverage (MCDC)

Major standards addressing code coverage:

DO‑178B (aviation): Requires 100% coverage for safety‑critical software.

IEC 61508 (functional safety): Recommends 100% coverage for certain metrics, with strictness depending on safety integrity level.

ISO 26262 (automotive): Requires coverage measurement and justification when coverage is insufficient; suggests different metrics for unit and integration testing.

ANSI/IEEE 1008‑1987 : Mandates 100% statement coverage and recommends 100% branch coverage for critical code.

FDA guidance for medical devices : Calls for structural coverage proportional to software risk, without specifying exact percentages.

These standards define coverage requirements in the context of risk analysis, safety integrity levels, and functional safety, but often leave the exact metric definitions to the implementer.

References include studies on efficient use of code coverage, its relation to quality, empirical analyses of branch coverage, and guidance on testing completeness.