diff --git a/.github/workflows/test-php.yml b/.github/workflows/test-php.yml
index b6daa6997..6a5bd41c3 100644
--- a/.github/workflows/test-php.yml
+++ b/.github/workflows/test-php.yml
@@ -28,6 +28,7 @@ jobs:
         uses: shivammathur/setup-php@v2
         with:
           php-version: "${{ matrix.php }}"
+          ini-values: "memory_limit=-1"
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
 
@@ -56,8 +57,23 @@ jobs:
         run: |
           vendor/bin/php-cs-fixer --dry-run --diff fix
           vendor/bin/psalm --no-cache
-          vendor/bin/phpunit
+          vendor/bin/phpunit --testsuite unit
 
-      - name: run acceptance tests
+      - name: Run acceptance tests
         run: make acceptance
         working-directory: php
+
+      - name: Mutation tests - minimum thresholds
+        run: |
+          vendor/bin/roave-infection-static-analysis-plugin \
+            --min-msi=90 \
+            --min-covered-msi=90
+        working-directory: php
+
+      - name: Mutation tests - modifications
+        run: |
+          git fetch --depth=1 origin $GITHUB_BASE_REF
+          vendor/bin/roave-infection-static-analysis-plugin \
+            --git-diff-lines --git-diff-base=origin/$GITHUB_BASE_REF \
+            --logger-github --ignore-msi-with-no-mutations
+        working-directory: php
diff --git a/php/.gitignore b/php/.gitignore
index 4a6f7c9af..f4467a9d5 100644
--- a/php/.gitignore
+++ b/php/.gitignore
@@ -1,5 +1,4 @@
 build
-acceptance
 vendor
 composer.lock
 .phpunit.cache
diff --git a/php/Makefile b/php/Makefile
index 034aa6b59..526b54913 100644
--- a/php/Makefile
+++ b/php/Makefile
@@ -6,16 +6,6 @@ GHERKIN_RAZOR = gherkin-php.razor
 SOURCE_FILES = $(shell find . -name "*.php"  | grep -v $(GHERKIN_PARSER))
 
 GHERKIN = bin/gherkin
-GHERKIN_GENERATE_TOKENS = bin/gherkin-generate-tokens
-
-GOOD_FEATURE_FILES = $(shell find ../testdata/good -name "*.feature")
-BAD_FEATURE_FILES  = $(shell find ../testdata/bad -name "*.feature")
-
-TOKENS       = $(patsubst ../testdata/%,acceptance/testdata/%.tokens,$(GOOD_FEATURE_FILES))
-ASTS         = $(patsubst ../testdata/%,acceptance/testdata/%.ast.ndjson,$(GOOD_FEATURE_FILES))
-PICKLES      = $(patsubst ../testdata/%,acceptance/testdata/%.pickles.ndjson,$(GOOD_FEATURE_FILES))
-SOURCES      = $(patsubst ../testdata/%,acceptance/testdata/%.source.ndjson,$(GOOD_FEATURE_FILES))
-ERRORS       = $(patsubst ../testdata/%,acceptance/testdata/%.errors.ndjson,$(BAD_FEATURE_FILES))
 
 .DEFAULT_GOAL = help
 
@@ -39,12 +29,13 @@ clean: ## Remove all build artifacts and files generated by the acceptance tests
 
 .DELETE_ON_ERROR:
 
-acceptance: .built $(TOKENS) $(ASTS) $(PICKLES) $(ERRORS) $(SOURCES) ## Build acceptance test dir and compare results with reference
+acceptance: .built  ## Test parser against test data
+	vendor/bin/phpunit --testsuite acceptance
 
-.built: vendor $(SOURCE_FILES)
+.built: vendor/autoload.php $(SOURCE_FILES)
 	touch $@
 
-vendor: composer.json
+vendor/autoload.php: composer.json
 	composer update
 
 $(GHERKIN_PARSER): $(GHERKIN_RAZOR) ../gherkin.berp
@@ -57,28 +48,3 @@ $(GHERKIN_PARSER): $(GHERKIN_RAZOR) ../gherkin.berp
 
 $(GHERKIN_LANGUAGES_JSON):
 	cp ../gherkin-languages.json $@
-
-acceptance/testdata/%.tokens: ../testdata/% ../testdata/%.tokens
-	mkdir -p $(@D)
-	$(GHERKIN_GENERATE_TOKENS) $< > $@
-	diff --unified $<.tokens $@
-
-acceptance/testdata/%.ast.ndjson: ../testdata/% ../testdata/%.ast.ndjson
-	mkdir -p $(@D)
-	$(GHERKIN) --no-source --no-pickles --predictable-ids $< | jq --sort-keys --compact-output "." > $@
-	diff --unified <(jq "." $<.ast.ndjson) <(jq "." $@)
-
-acceptance/testdata/%.pickles.ndjson: ../testdata/% ../testdata/%.pickles.ndjson
-	mkdir -p $(@D)
-	$(GHERKIN) --no-source --no-ast --predictable-ids $< | jq --sort-keys --compact-output "." > $@
-	diff --unified <(jq "." $<.pickles.ndjson) <(jq "." $@)
-
-acceptance/testdata/%.source.ndjson: ../testdata/% ../testdata/%.source.ndjson
-	mkdir -p $(@D)
-	$(GHERKIN) --no-ast --no-pickles --predictable-ids $< | jq --sort-keys --compact-output "." > $@
-	diff --unified <(jq "." $<.source.ndjson) <(jq "." $@)
-
-acceptance/testdata/%.errors.ndjson: ../testdata/% ../testdata/%.errors.ndjson
-	mkdir -p $(@D)
-	$(GHERKIN) --no-source --predictable-ids $< | jq --sort-keys --compact-output "." > $@
-	diff --unified <(jq "." $<.errors.ndjson) <(jq "." $@)
diff --git a/php/bin/gherkin-generate-tokens b/php/bin/gherkin-generate-tokens
deleted file mode 100755
index 0c9ce2ed8..000000000
--- a/php/bin/gherkin-generate-tokens
+++ /dev/null
@@ -1,24 +0,0 @@
-#!/usr/bin/env php
-<?php declare(strict_types=1);
-
-/**
- * Script that outputs debug tokens based on a list of files
- */
-
-namespace Cucumber\Gherkin;
-
-require __DIR__ . '/../vendor/autoload.php';
-
-$parser = new Parser(new TokenFormatterBuilder());
-
-// first element is the script name
-array_shift($argv);
-
-foreach($argv as $fileName) {
-    $result = $parser->parse(
-            $fileName,
-            new StringTokenScanner(file_get_contents($fileName)),
-            new TokenMatcher(),
-    );
-    echo $result;
-}
diff --git a/php/composer.json b/php/composer.json
index 5452d158a..302750490 100644
--- a/php/composer.json
+++ b/php/composer.json
@@ -22,7 +22,9 @@
         "vimeo/psalm": "^4.24",
         "friendsofphp/php-cs-fixer": "^3.5",
         "psalm/plugin-phpunit": "^0.18.0",
-        "nikic/php-parser": "^4.14"
+        "nikic/php-parser": "^4.14",
+        "infection/infection": "^0.26.16",
+        "roave/infection-static-analysis-plugin": "^1.25"
     },
     "repositories": [
         {
@@ -34,5 +36,10 @@
                 }
             }
         }
-    ]
+    ],
+    "config": {
+        "allow-plugins": {
+            "infection/extension-installer": false
+        }
+    }
 }
diff --git a/php/infection.json5 b/php/infection.json5
new file mode 100644
index 000000000..6c187660f
--- /dev/null
+++ b/php/infection.json5
@@ -0,0 +1,11 @@
+{
+    "$schema": "vendor/infection/infection/resources/schema.json",
+    "source": {
+        "directories": [
+            "src",
+        ]
+    },
+    "mutators": {
+        "@default": true
+    }
+}
diff --git a/php/phpunit.xml b/php/phpunit.xml
index fd5e5582f..7e569f1af 100644
--- a/php/phpunit.xml
+++ b/php/phpunit.xml
@@ -12,8 +12,11 @@
          failOnWarning="true"
          verbose="true">
     <testsuites>
-        <testsuite name="default">
-            <directory>tests</directory>
+        <testsuite name="unit">
+            <directory>tests/unit</directory>
+        </testsuite>
+        <testsuite name="acceptance">
+            <file>tests/acceptance/TestDataTest.php</file>
         </testsuite>
     </testsuites>
 
diff --git a/php/psalm.xml b/php/psalm.xml
index 6686ed61d..f267e4ebb 100644
--- a/php/psalm.xml
+++ b/php/psalm.xml
@@ -7,7 +7,6 @@
     xsi:schemaLocation="https://getpsalm.org/schema/config vendor/vimeo/psalm/config.xsd"
 >
     <projectFiles>
-        <file name="bin/gherkin-generate-tokens"/>
         <file name="bin/gherkin"/>
         <directory name="src"/>
         <directory name="src-generated"/>
diff --git a/php/tests/acceptance/TestDataTest.php b/php/tests/acceptance/TestDataTest.php
new file mode 100644
index 000000000..0eaa7b607
--- /dev/null
+++ b/php/tests/acceptance/TestDataTest.php
@@ -0,0 +1,144 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Cucumber\Gherkin;
+
+use Cucumber\Messages\Envelope;
+use Cucumber\Messages\Source;
+use Cucumber\Messages\Streams\NdJson\NdJsonStreamWriter;
+use PHPUnit\Framework\TestCase;
+
+final class TestDataTest extends TestCase
+{
+    /**
+     * @dataProvider provideGoodFeatureFiles
+     */
+    public function testTokensAreSameAsTestData(string $fullPath): void
+    {
+        $result = (new Parser(new TokenFormatterBuilder()))->parse(
+            $fullPath,
+            new StringTokenScanner(file_get_contents($fullPath)),
+            new TokenMatcher(),
+        );
+
+        self::assertStringEqualsFile($fullPath . '.tokens', $result);
+    }
+
+    /**
+     * @dataProvider provideGoodFeatureFiles
+     */
+    public function testAstsAreSameAsTestData(string $fullPath, Source $source): void
+    {
+        $envelopes = (new GherkinParser(
+            predictableIds: true,
+            includeSource: false,
+            includeGherkinDocument: true,
+            includePickles: false,
+        ))->parse([$source]);
+
+        self::assertEnvelopesMatchNdJsonFile($envelopes, $fullPath . '.ast.ndjson');
+    }
+
+    /**
+     * @dataProvider provideGoodFeatureFiles
+     */
+    public function testSourcesAreSameAsTestData(string $fullPath, Source $source): void
+    {
+        $envelopes = (new GherkinParser(
+            predictableIds: true,
+            includeSource: true,
+            includeGherkinDocument: false,
+            includePickles: false,
+        ))->parse([$source]);
+
+        self::assertEnvelopesMatchNdJsonFile($envelopes, $fullPath . '.source.ndjson');
+    }
+
+    /**
+     * @dataProvider provideGoodFeatureFiles
+     */
+    public function testPicklesAreSameAsTestData(string $fullPath, Source $source): void
+    {
+        $envelopes = (new GherkinParser(
+            predictableIds: true,
+            includeSource: false,
+            includeGherkinDocument: false,
+            includePickles: true,
+        ))->parse([$source]);
+
+        self::assertEnvelopesMatchNdJsonFile($envelopes, $fullPath . '.pickles.ndjson');
+    }
+
+    /**
+     * @dataProvider provideBadFeatureFiles
+     */
+    public function testErrorsAreSameAsTestData(string $fullPath, Source $source): void
+    {
+        $envelopes = (new GherkinParser(
+            predictableIds: true,
+            includeSource: false,
+            includeGherkinDocument: false,
+            includePickles: true,
+        ))->parse([$source]);
+
+        self::assertEnvelopesMatchNdJsonFile($envelopes, $fullPath . '.errors.ndjson');
+    }
+
+    /**
+     * @return iterable<string, array{0: string, 1: Source}>
+     */
+    public function provideGoodFeatureFiles(): iterable
+    {
+        return $this->provideFeatureFiles("good");
+    }
+
+    /**
+     * @return iterable<string, array{0: string, 1: Source}>
+     */
+    public function provideBadFeatureFiles(): iterable
+    {
+        return $this->provideFeatureFiles("bad");
+    }
+
+    /**
+     * @param 'good'|'bad' $subDir
+     *
+     * @return iterable<string, array{0: string, 1: Source}>
+     */
+    private function provideFeatureFiles(string $subDir): iterable
+    {
+        foreach (glob(__DIR__ . "/../../../testdata/$subDir/*.feature") as $fullPath) {
+            $shortPath = substr($fullPath, strlen(__DIR__ . '/../../'));
+
+            yield $shortPath => [$fullPath, new Source($shortPath, file_get_contents($fullPath))];
+        }
+    }
+
+    /**
+     * @param iterable<Envelope> $envelopes
+     */
+    private static function assertEnvelopesMatchNdJsonFile(iterable $envelopes, string $expectedfile): void
+    {
+        $output = fopen('php://memory', 'w');
+        NdJsonStreamWriter::fromFileHandle($output)->writeEnvelopes($envelopes);
+        rewind($output);
+
+        $actual = stream_get_contents($output);
+        $expected = file_get_contents($expectedfile);
+
+        // rather than compare the full file, compare line by line to get better JSON diffs on error
+        $actualLines = explode("\n", $actual);
+        $expectedLines = explode("\n", $expected);
+
+        self::assertSame(count($actualLines), count($expectedLines));
+
+        foreach ($actualLines as $i => $actualLine) {
+            if ($actualLine !== '') {
+                self::assertJsonStringEqualsJsonString($expectedLines[$i], $actualLine);
+            } else {
+                self::assertEquals($expectedLines[$i], '');
+            }
+        }
+    }
+}
diff --git a/php/tests/AstNodeTest.php b/php/tests/unit/AstNodeTest.php
similarity index 100%
rename from php/tests/AstNodeTest.php
rename to php/tests/unit/AstNodeTest.php
diff --git a/php/tests/GherkinDialectProviderTest.php b/php/tests/unit/GherkinDialectProviderTest.php
similarity index 100%
rename from php/tests/GherkinDialectProviderTest.php
rename to php/tests/unit/GherkinDialectProviderTest.php
diff --git a/php/tests/GherkinDialectTest.php b/php/tests/unit/GherkinDialectTest.php
similarity index 100%
rename from php/tests/GherkinDialectTest.php
rename to php/tests/unit/GherkinDialectTest.php
diff --git a/php/tests/GherkinLineSpanTest.php b/php/tests/unit/GherkinLineSpanTest.php
similarity index 100%
rename from php/tests/GherkinLineSpanTest.php
rename to php/tests/unit/GherkinLineSpanTest.php
diff --git a/php/tests/Parser/RuleTypeTest.php b/php/tests/unit/Parser/RuleTypeTest.php
similarity index 100%
rename from php/tests/Parser/RuleTypeTest.php
rename to php/tests/unit/Parser/RuleTypeTest.php
diff --git a/php/tests/ParserException/CompositeParserExceptionTest.php b/php/tests/unit/ParserException/CompositeParserExceptionTest.php
similarity index 100%
rename from php/tests/ParserException/CompositeParserExceptionTest.php
rename to php/tests/unit/ParserException/CompositeParserExceptionTest.php
diff --git a/php/tests/ParserException/NoSuchLanguageExceptionTest.php b/php/tests/unit/ParserException/NoSuchLanguageExceptionTest.php
similarity index 100%
rename from php/tests/ParserException/NoSuchLanguageExceptionTest.php
rename to php/tests/unit/ParserException/NoSuchLanguageExceptionTest.php
diff --git a/php/tests/ParserException/UnexpectedEofExceptionTest.php b/php/tests/unit/ParserException/UnexpectedEofExceptionTest.php
similarity index 100%
rename from php/tests/ParserException/UnexpectedEofExceptionTest.php
rename to php/tests/unit/ParserException/UnexpectedEofExceptionTest.php
diff --git a/php/tests/ParserException/UnexpectedTokenExceptionTest.php b/php/tests/unit/ParserException/UnexpectedTokenExceptionTest.php
similarity index 100%
rename from php/tests/ParserException/UnexpectedTokenExceptionTest.php
rename to php/tests/unit/ParserException/UnexpectedTokenExceptionTest.php
diff --git a/php/tests/StringGherkinLineTest.php b/php/tests/unit/StringGherkinLineTest.php
similarity index 100%
rename from php/tests/StringGherkinLineTest.php
rename to php/tests/unit/StringGherkinLineTest.php
diff --git a/php/tests/StringTokenScannerTest.php b/php/tests/unit/StringTokenScannerTest.php
similarity index 100%
rename from php/tests/StringTokenScannerTest.php
rename to php/tests/unit/StringTokenScannerTest.php
diff --git a/php/tests/StringUtilsTest.php b/php/tests/unit/StringUtilsTest.php
similarity index 100%
rename from php/tests/StringUtilsTest.php
rename to php/tests/unit/StringUtilsTest.php
diff --git a/php/tests/TokenFormatterBuilderTest.php b/php/tests/unit/TokenFormatterBuilderTest.php
similarity index 100%
rename from php/tests/TokenFormatterBuilderTest.php
rename to php/tests/unit/TokenFormatterBuilderTest.php
diff --git a/php/tests/TokenFormatterTest.php b/php/tests/unit/TokenFormatterTest.php
similarity index 100%
rename from php/tests/TokenFormatterTest.php
rename to php/tests/unit/TokenFormatterTest.php
diff --git a/php/tests/TokenMatchTest.php b/php/tests/unit/TokenMatchTest.php
similarity index 100%
rename from php/tests/TokenMatchTest.php
rename to php/tests/unit/TokenMatchTest.php
diff --git a/php/tests/TokenMatcherTest.php b/php/tests/unit/TokenMatcherTest.php
similarity index 100%
rename from php/tests/TokenMatcherTest.php
rename to php/tests/unit/TokenMatcherTest.php
diff --git a/php/tests/TokenTest.php b/php/tests/unit/TokenTest.php
similarity index 100%
rename from php/tests/TokenTest.php
rename to php/tests/unit/TokenTest.php