Introduction to mutation testing

Code coverage is an useful metric tests definition. But it's not the single one metric which can be used. One alternative are mutation tests.

Looking for a better data engineering position and skills?

You have been working as a data engineer but feel stuck? You don't have any new challenges and are still writing the same jobs all over again? You have now different options. You can try to look for a new job, now or later, or learn from the others! "Become a Better Data Engineer" initiative is one of these places where you can find online learning resources where the theory meets the practice. They will help you prepare maybe for the next job, or at least, improve your current skillset without looking for something else.

👉 I'm interested in improving my data engineering skillset

See you there, Bartosz

The post is divided in 2 parts. The first one defines mutation tests. The second shows how to implement them in Java.


Mutation tests are a technique responding to the philosophical problem of "Who will guard the guards ?". In testing world it's translated by "Who will check if the tests test correct stuff ?". In other words, mutation test check if written tests will fail when tested code will mutate. The mutation can be represented as for example the change of boundary value such as MIN_LENGTH in below code:

public boolean isCorrectName(String name) {
  return name != null && name.length() > MIN_LENGTH;

After changing MIN_LENGTH mutation test is expected to fail. If it does, it's considered as correct. Otherwise, it indicates that it's incorrectly defined and for example it tests bad piece of code.

We distinguish 3 types of mutation tests:

Mutation test lifecycle

Mutation test follows a strict lifecycle where the word mutant appears plenty of times. This word represents the code containing mutation, i.e. modified with the goal to fail the test. As you can imagine, the first step of mutation test is the creation of mutant.

Once all mutants are introduced to code, the original tests are executed against mutated code. If the test fails, the mutant is killed. It means that mutation test passes. In the other side, when the original test doesn't fail, the mutant is alive.

Killing the mutant is a synonym of 3 conditions to met, commonly called RIP model:

  1. Mutation statement must be reached by the original test.
  2. Test input must produce different behavior on mutated and not mutated tested code.
  3. Incorrect state must be checked by the test.

The result of mutation tests is represented as mutation score, calculated from given formula:

MutationScore = KilledMutants / AllMutants

Mutation test example

We'll see an example of mutation test thanks to PitTest framework and its IntelliJ plugin. A pretty simple example will be used:

public class MutationTest {

  private DecisionMaker decisionMaker = new DecisionMaker();

  public void should_detect_4_letters_name_as_valid() {

  public void should_detect_user_as_minor() {


public class DecisionMaker {

  private static final int MIN_LENGTH = 3;

  private static final int MAJOR_AGE = 18;

  public boolean isNameValid(String name) {
    return name != null && name.length() > MIN_LENGTH;

  public boolean isMajor(int age) {
    return age >= MAJOR_AGE;

The configuration used to ran this code through the Pit plugin is --outputFormats HTML --verbose true --exportLineCoverage true --targetTests com.waitingforcode.tdd.mutation.* . After executing the the first time, you'll see that 2 test cases are missing for boundary conditions. The first is the check on 3-letters name. The second corresponds to the age greater or equal to 18:

Thus, we add 2 tests to cover these conditions:

public void should_detect_3_letters_as_invalid() {

public void should_detect_user_as_major() {

Now, after re-executing Pit, we'll see that all test cases are covered (= all mutants are killed):

Through this post we can see how much mutation tests can help in the definition of well covered tests. The first part presents the idea of mutation generally. We can learn that mutations tests are a kind of "test guards" detecting if written tests are correct and exhaustive.