diff --git a/modules/java-http-client/test/src/java/net/http/HttpResponse/BodySubscribersLineTest.scala b/modules/java-http-client/test/src/java/net/http/HttpResponse/BodySubscribersLineTest.scala
new file mode 100644
index 0000000..c0356d2
--- /dev/null
+++ b/modules/java-http-client/test/src/java/net/http/HttpResponse/BodySubscribersLineTest.scala
@@ -0,0 +1,119 @@
+package snhttp.test.java.net.http
+
+import java.net.http.HttpResponse.BodySubscribers
+import java.nio.ByteBuffer
+import java.nio.charset.StandardCharsets.UTF_8
+import java.util.List as JList
+import java.util.concurrent.TimeUnit
+import java.util.function.Function
+
+import utest.{TestSuite, Tests, test, assert}
+
+import _root_.snhttp.test.jdk.net.http.{MockSubscriber, MockSubscription}
+
+class BodySubscribersLineTest extends TestSuite:
+
+  private def checkLines(
+      events: Seq[Seq[Array[Byte]]],
+      separator: String,
+      expected: Seq[String],
+  ): Unit = {
+    val downstream = MockSubscriber[String](request = true)
+    val finisher: Function[MockSubscriber[String], Int] = _.received.size
+    val subscriber = BodySubscribers.fromLineSubscriber(downstream, finisher, UTF_8, separator)
+
+    subscriber.onSubscribe(MockSubscription())
+    events.foreach { chunks =>
+      val buffers = new java.util.ArrayList[ByteBuffer]()
+      chunks.foreach(chunk => (buffers.add(ByteBuffer.wrap(chunk)): Unit))
+      subscriber.onNext(buffers)
+    }
+    subscriber.onComplete()
+
+    val count = subscriber.getBody().toCompletableFuture().get(5, TimeUnit.SECONDS)
+    assert(count == expected.size)
+    assert(downstream.received.toSeq == expected)
+    assert(downstream.completes == 1)
+    assert(downstream.errors == 0)
+  }
+
+  val tests = Tests:
+
+    test("UTF-8 characters split between buffers in one onNext") {
+      // Exercise every split, including inside two-, three-, and four-byte characters.
+      val text = "a\u00a2\u20ac\ud83d\ude00z\nlast"
+      val bytes = text.getBytes(UTF_8)
+      for split <- 1 until bytes.length do
+        checkLines(
+          Seq(Seq(bytes.take(split), bytes.drop(split))),
+          "\n",
+          Seq("a\u00a2\u20ac\ud83d\ude00z", "last"),
+        )
+    }
+
+    test("UTF-8 characters split between onNext calls") {
+      val text = "a\u00a2\u20ac\ud83d\ude00z\nlast"
+      val bytes = text.getBytes(UTF_8)
+      for split <- 1 until bytes.length do
+        checkLines(
+          Seq(Seq(bytes.take(split)), Seq(Array.emptyByteArray), Seq(bytes.drop(split))),
+          "\n",
+          Seq("a\u00a2\u20ac\ud83d\ude00z", "last"),
+        )
+    }
+
+    test("UTF-8 characters delivered one byte at a time") {
+      val text = "\u00a2\u20ac\ud83d\ude00\n"
+      checkLines(
+        text.getBytes(UTF_8).toSeq.map(byte => Seq(Array(byte))),
+        "\n",
+        Seq("\u00a2\u20ac\ud83d\ude00"),
+      )
+    }
+
+    test("null separator recognizes CR LF and CRLF") {
+      checkLines(
+        Seq(Seq("first\rsecond\nthird\r\n\nlast\r".getBytes(UTF_8))),
+        null,
+        Seq("first", "second", "third", "", "last"),
+      )
+    }
+
+    test("one-argument overload recognizes CR LF and CRLF") {
+      val downstream = MockSubscriber[String](request = true)
+      val subscriber = BodySubscribers.fromLineSubscriber(downstream)
+      subscriber.onSubscribe(MockSubscription())
+      subscriber.onNext(JList.of(ByteBuffer.wrap("a\rb\nc\r\nd".getBytes(UTF_8))))
+      subscriber.onComplete()
+
+      assert(subscriber.getBody().toCompletableFuture().get(5, TimeUnit.SECONDS) == null)
+      assert(downstream.received.toSeq == Seq("a", "b", "c", "d"))
+      assert(downstream.completes == 1)
+      assert(downstream.errors == 0)
+    }
+
+    test("default CRLF can span events without creating an extra line") {
+      checkLines(
+        Seq("first\r", "", "\nsecond\r", "third\n", "tail").map(s => Seq(s.getBytes(UTF_8))),
+        null,
+        Seq("first", "second", "third", "tail"),
+      )
+    }
+
+    test("explicit LF separator preserves carriage returns") {
+      checkLines(
+        Seq("a\rb\nc\r", "\ntail").map(s => Seq(s.getBytes(UTF_8))),
+        "\n",
+        Seq("a\rb", "c\r", "tail"),
+      )
+    }
+
+    test("incomplete UTF-8 sequence is replaced at end of input") {
+      checkLines(
+        Seq(Seq(Array(0xe2.toByte)), Seq(Array(0x82.toByte))),
+        "\n",
+        Seq("\ufffd"),
+      )
+    }
+
+end BodySubscribersLineTest