diff --git a/doc/api/stream.md b/doc/api/stream.md
index 6ee016466206d8..25a218cf130728 100644
--- a/doc/api/stream.md
+++ b/doc/api/stream.md
@@ -437,6 +437,14 @@ process.nextTick(() => {
 
 See also: [`writable.cork()`][].
 
+##### writable.writableHighWaterMark
+<!-- YAML
+added: REPLACEME
+-->
+
+Return the value of `highWaterMark` passed when constructing this
+`Writable`.
+
 ##### writable.write(chunk[, encoding][, callback])
 <!-- YAML
 added: v0.9.4
@@ -879,6 +887,14 @@ to prevent memory leaks.
 never closed until the Node.js process exits, regardless of the specified
 options.
 
+##### readable.readableHighWaterMark
+<!-- YAML
+added: REPLACEME
+-->
+
+Return the value of `highWaterMark` passed when constructing this
+`Readable`.
+
 ##### readable.read([size])
 <!-- YAML
 added: v0.9.4
diff --git a/lib/_http_server.js b/lib/_http_server.js
index 8fe9e58dff5560..736762dc26c4dc 100644
--- a/lib/_http_server.js
+++ b/lib/_http_server.js
@@ -373,13 +373,13 @@ function connectionListener(socket) {
 function updateOutgoingData(socket, state, delta) {
   state.outgoingData += delta;
   if (socket._paused &&
-      state.outgoingData < socket._writableState.highWaterMark) {
+      state.outgoingData < socket.writeHWM) {
     return socketOnDrain(socket, state);
   }
 }
 
 function socketOnDrain(socket, state) {
-  var needPause = state.outgoingData > socket._writableState.highWaterMark;
+  var needPause = state.outgoingData > socket.writeHWM;
 
   // If we previously paused, then start reading again.
   if (socket._paused && !needPause) {
@@ -569,7 +569,7 @@ function parserOnIncoming(server, socket, state, req, keepAlive) {
   // pipelined requests that may never be resolved.
   if (!socket._paused) {
     var ws = socket._writableState;
-    if (ws.needDrain || state.outgoingData >= ws.highWaterMark) {
+    if (ws.needDrain || state.outgoingData >= socket.writableHighWaterMark) {
       socket._paused = true;
       // We also need to pause the parser, but don't do that until after
       // the call to execute, because we may still be processing the last
diff --git a/lib/_stream_duplex.js b/lib/_stream_duplex.js
index 7440cd08729e1c..05f649340845b0 100644
--- a/lib/_stream_duplex.js
+++ b/lib/_stream_duplex.js
@@ -34,11 +34,14 @@ const Writable = require('_stream_writable');
 
 util.inherits(Duplex, Readable);
 
-var keys = Object.keys(Writable.prototype);
-for (var v = 0; v < keys.length; v++) {
-  var method = keys[v];
-  if (!Duplex.prototype[method])
-    Duplex.prototype[method] = Writable.prototype[method];
+{
+  // avoid scope creep, the keys array can then be collected
+  const keys = Object.keys(Writable.prototype);
+  for (var v = 0; v < keys.length; v++) {
+    const method = keys[v];
+    if (!Duplex.prototype[method])
+      Duplex.prototype[method] = Writable.prototype[method];
+  }
 }
 
 function Duplex(options) {
@@ -61,6 +64,16 @@ function Duplex(options) {
   this.once('end', onend);
 }
 
+Object.defineProperty(Duplex.prototype, 'writableHighWaterMark', {
+  // making it explicit this property is not enumerable
+  // because otherwise some prototype manipulation in
+  // userland will fail
+  enumerable: false,
+  get: function() {
+    return this._writableState.highWaterMark;
+  }
+});
+
 // the no-half-open enforcer
 function onend() {
   // if we allow half-open state, or if the writable side ended,
diff --git a/lib/_stream_readable.js b/lib/_stream_readable.js
index 6427d2e5f6cb49..452b7321b2c480 100644
--- a/lib/_stream_readable.js
+++ b/lib/_stream_readable.js
@@ -916,6 +916,15 @@ Readable.prototype.wrap = function(stream) {
   return self;
 };
 
+Object.defineProperty(Readable.prototype, 'readableHighWaterMark', {
+  // making it explicit this property is not enumerable
+  // because otherwise some prototype manipulation in
+  // userland will fail
+  enumerable: false,
+  get: function() {
+    return this._readableState.highWaterMark;
+  }
+});
 
 // exposed for testing purposes only.
 Readable._fromList = fromList;
diff --git a/lib/_stream_writable.js b/lib/_stream_writable.js
index 13b79233e91bfb..1735fafb5f34d3 100644
--- a/lib/_stream_writable.js
+++ b/lib/_stream_writable.js
@@ -334,6 +334,16 @@ function decodeChunk(state, chunk, encoding) {
   return chunk;
 }
 
+Object.defineProperty(Writable.prototype, 'writableHighWaterMark', {
+  // making it explicit this property is not enumerable
+  // because otherwise some prototype manipulation in
+  // userland will fail
+  enumerable: false,
+  get: function() {
+    return this._writableState.highWaterMark;
+  }
+});
+
 // if we're already writing something, then just put this
 // in the queue, and wait our turn.  Otherwise, call _write
 // If we return false, then we need a drain event, so set that flag.
diff --git a/lib/fs.js b/lib/fs.js
index 215acf9b2436f0..01edc7411a5c02 100644
--- a/lib/fs.js
+++ b/lib/fs.js
@@ -2072,7 +2072,7 @@ ReadStream.prototype._read = function(n) {
 
   if (!pool || pool.length - pool.used < kMinPoolSpace) {
     // discard the old pool.
-    allocNewPool(this._readableState.highWaterMark);
+    allocNewPool(this.readableHighWaterMark);
   }
 
   // Grab another reference to the pool in the case that while we're
diff --git a/test/parallel/test-http-pipeline-regr-3508.js b/test/parallel/test-http-pipeline-regr-3508.js
index ff735818f68121..02c1e1787be6f3 100644
--- a/test/parallel/test-http-pipeline-regr-3508.js
+++ b/test/parallel/test-http-pipeline-regr-3508.js
@@ -30,7 +30,7 @@ const server = http.createServer(function(req, res) {
     res.end(chunk);
   }
   size += res.outputSize;
-  if (size <= req.socket._writableState.highWaterMark) {
+  if (size <= req.socket.writableHighWaterMark) {
     more();
     return;
   }
diff --git a/test/parallel/test-stream-big-packet.js b/test/parallel/test-stream-big-packet.js
index 18b9b7ee4fc348..cf116b438909c3 100644
--- a/test/parallel/test-stream-big-packet.js
+++ b/test/parallel/test-stream-big-packet.js
@@ -51,7 +51,7 @@ s1.pipe(s3);
 s2.pipe(s3, { end: false });
 
 // We must write a buffer larger than highWaterMark
-const big = Buffer.alloc(s1._writableState.highWaterMark + 1, 'x');
+const big = Buffer.alloc(s1.writableHighWaterMark + 1, 'x');
 
 // Since big is larger than highWaterMark, it will be buffered internally.
 assert(!s1.write(big));
diff --git a/test/parallel/test-stream-readable-flow-recursion.js b/test/parallel/test-stream-readable-flow-recursion.js
index 4dcec8901f9359..744298946c5eb1 100644
--- a/test/parallel/test-stream-readable-flow-recursion.js
+++ b/test/parallel/test-stream-readable-flow-recursion.js
@@ -68,7 +68,7 @@ flow(stream, 5000, function() {
 process.on('exit', function(code) {
   assert.strictEqual(reads, 2);
   // we pushed up the high water mark
-  assert.strictEqual(stream._readableState.highWaterMark, 8192);
+  assert.strictEqual(stream.readableHighWaterMark, 8192);
   // length is 0 right now, because we pulled it all out.
   assert.strictEqual(stream._readableState.length, 0);
   assert(!code);
diff --git a/test/parallel/test-stream-transform-split-objectmode.js b/test/parallel/test-stream-transform-split-objectmode.js
index 6d9603fca37909..55d613573493b9 100644
--- a/test/parallel/test-stream-transform-split-objectmode.js
+++ b/test/parallel/test-stream-transform-split-objectmode.js
@@ -29,8 +29,12 @@ const parser = new Transform({ readableObjectMode: true });
 
 assert(parser._readableState.objectMode);
 assert(!parser._writableState.objectMode);
-assert.strictEqual(parser._readableState.highWaterMark, 16);
-assert.strictEqual(parser._writableState.highWaterMark, 16 * 1024);
+assert.strictEqual(parser.readableHighWaterMark, 16);
+assert.strictEqual(parser.writableHighWaterMark, 16 * 1024);
+assert.strictEqual(parser.readableHighWaterMark,
+                   parser._readableState.highWaterMark);
+assert.strictEqual(parser.writableHighWaterMark,
+                   parser._writableState.highWaterMark);
 
 parser._transform = function(chunk, enc, callback) {
   callback(null, { val: chunk[0] });
@@ -53,8 +57,12 @@ const serializer = new Transform({ writableObjectMode: true });
 
 assert(!serializer._readableState.objectMode);
 assert(serializer._writableState.objectMode);
-assert.strictEqual(serializer._readableState.highWaterMark, 16 * 1024);
-assert.strictEqual(serializer._writableState.highWaterMark, 16);
+assert.strictEqual(serializer.readableHighWaterMark, 16 * 1024);
+assert.strictEqual(serializer.writableHighWaterMark, 16);
+assert.strictEqual(parser.readableHighWaterMark,
+                   parser._readableState.highWaterMark);
+assert.strictEqual(parser.writableHighWaterMark,
+                   parser._writableState.highWaterMark);
 
 serializer._transform = function(obj, _, callback) {
   callback(null, Buffer.from([obj.val]));
diff --git a/test/parallel/test-stream2-unpipe-leak.js b/test/parallel/test-stream2-unpipe-leak.js
index 27cad3d5f575c6..cc331d58217e25 100644
--- a/test/parallel/test-stream2-unpipe-leak.js
+++ b/test/parallel/test-stream2-unpipe-leak.js
@@ -68,6 +68,6 @@ console.error(src._readableState);
 process.on('exit', function() {
   src._readableState.buffer.length = 0;
   console.error(src._readableState);
-  assert(src._readableState.length >= src._readableState.highWaterMark);
+  assert(src._readableState.length >= src.readableHighWaterMark);
   console.log('ok');
 });