diff --git a/docs/news.rst b/docs/news.rst
index 4b8440208..d9c95848e 100644
--- a/docs/news.rst
+++ b/docs/news.rst
@@ -34,6 +34,7 @@ Scrapy changes:
 - downloader handlers (:setting:`DOWNLOAD_HANDLERS` setting) now receive settings as the first argument of the constructor
 - replaced memory usage acounting with (more portable) `resource`_ module, removed ``scrapy.utils.memory`` module
 - removed signal: ``scrapy.mail.mail_sent``
+- removed ``TRACK_REFS`` setting, now :ref:`trackrefs <topics-leaks-trackrefs>` is always enabled
 
 Scrapyd changes:
 
diff --git a/docs/topics/leaks.rst b/docs/topics/leaks.rst
index 207cc36af..95a41f360 100644
--- a/docs/topics/leaks.rst
+++ b/docs/topics/leaks.rst
@@ -56,14 +56,9 @@ Debugging memory leaks with ``trackref``
 memory leaks. It basically tracks the references to all live Requests,
 Responses, Item and Selector objects. 
 
-To activate the ``trackref`` module, enable the :setting:`TRACK_REFS` setting.
-It only imposes a minor performance impact, so it should be OK to use it, even
-in production environments.
-
-Once you have ``trackref`` enabled, you can enter the telnet console and inspect
-how many objects (of the classes mentioned above) are currently alive using the
-``prefs()`` function which is an alias to the
-:func:`~scrapy.utils.trackref.print_live_refs` function::
+You can enter the telnet console and inspect how many objects (of the classes
+mentioned above) are currently alive using the ``prefs()`` function which is an
+alias to the :func:`~scrapy.utils.trackref.print_live_refs` function::
 
     telnet localhost 6023
 
diff --git a/scrapy/contrib/debug.py b/scrapy/contrib/debug.py
index d19786c13..18a746d31 100644
--- a/scrapy/contrib/debug.py
+++ b/scrapy/contrib/debug.py
@@ -19,7 +19,6 @@ class StackTraceDump(object):
 
     def __init__(self, crawler=None):
         self.crawler = crawler
-        self.dumprefs = crawler.settings.getbool('TRACK_REFS')
         try:
             signal.signal(signal.SIGUSR2, self.dump_stacktrace)
             signal.signal(signal.SIGQUIT, self.dump_stacktrace)
@@ -34,7 +33,7 @@ class StackTraceDump(object):
     def dump_stacktrace(self, signum, frame):
         stackdumps = self._thread_stacks()
         enginestatus = format_engine_status(self.crawler.engine)
-        liverefs = format_live_refs() if self.dumprefs else ""
+        liverefs = format_live_refs()
         msg = "Dumping stack trace and engine status" \
             "\n{0}\n{1}\n{2}".format(enginestatus, liverefs, stackdumps)
         log.msg(msg)
diff --git a/scrapy/contrib/memdebug.py b/scrapy/contrib/memdebug.py
index 8965a1ebd..359c7ec8a 100644
--- a/scrapy/contrib/memdebug.py
+++ b/scrapy/contrib/memdebug.py
@@ -12,20 +12,19 @@ from scrapy.utils.trackref import live_refs
 
 class MemoryDebugger(object):
 
-    def __init__(self, stats, trackrefs=False):
+    def __init__(self, stats):
         try:
             import libxml2
             self.libxml2 = libxml2
         except ImportError:
             self.libxml2 = None
         self.stats = stats
-        self.trackrefs = trackrefs
 
     @classmethod
     def from_crawler(cls, crawler):
         if not crawler.settings.getbool('MEMDEBUG_ENABLED'):
             raise NotConfigured
-        o = cls(crawler.stats, crawler.settings.getbool('TRACK_REFS'))
+        o = cls(crawler.stats)
         crawler.signals.connect(o.engine_started, signals.engine_started)
         crawler.signals.connect(o.engine_stopped, signals.engine_stopped)
         return o
@@ -40,8 +39,7 @@ class MemoryDebugger(object):
             self.stats.set_value('memdebug/libxml2_leaked_bytes', self.libxml2.debugMemory(1))
         gc.collect()
         self.stats.set_value('memdebug/gc_garbage_count', len(gc.garbage))
-        if self.trackrefs:
-            for cls, wdict in live_refs.iteritems():
-                if not wdict:
-                    continue
-                self.stats.set_value('memdebug/live_refs/%s' % cls.__name__, len(wdict))
+        for cls, wdict in live_refs.iteritems():
+            if not wdict:
+                continue
+            self.stats.set_value('memdebug/live_refs/%s' % cls.__name__, len(wdict))
diff --git a/scrapy/utils/trackref.py b/scrapy/utils/trackref.py
index e28489ff3..8f3110f0a 100644
--- a/scrapy/utils/trackref.py
+++ b/scrapy/utils/trackref.py
@@ -2,8 +2,7 @@
 references to live object instances.
 
 If you want live objects for a particular class to be tracked, you only have to
-subclass form object_ref (instead of object). Also, remember to turn on
-tracking by enabling the TRACK_REFS setting.
+subclass form object_ref (instead of object).
 
 About performance: This library has a minimal performance impact when enabled,
 and no performance penalty at all when disabled (as object_ref becomes just an
@@ -16,8 +15,6 @@ from time import time
 from operator import itemgetter
 from types import NoneType
 
-from scrapy.conf import settings
-
 live_refs = defaultdict(weakref.WeakKeyDictionary)
 
 class object_ref(object):
@@ -31,12 +28,7 @@ class object_ref(object):
         live_refs[cls][obj] = time()
         return obj
 
-if not settings.getbool('TRACK_REFS'):
-    object_ref = object
-
 def format_live_refs(ignore=NoneType):
-    if object_ref is object:
-        return "The trackref module is disabled. Use TRACK_REFS setting to enable it."
     s = "Live References" + os.linesep + os.linesep
     now = time()
     for cls, wdict in live_refs.iteritems():