diff --git a/discord/ext/commands/cooldowns.py b/discord/ext/commands/cooldowns.py
index 5e7f2aa39..fe763fb62 100644
--- a/discord/ext/commands/cooldowns.py
+++ b/discord/ext/commands/cooldowns.py
@@ -26,13 +26,17 @@ DEALINGS IN THE SOFTWARE.
 
 from discord.enums import Enum
 import time
+import asyncio
+from collections import deque
 
 from ...abc import PrivateChannel
+from .errors import MaxConcurrencyReached
 
 __all__ = (
     'BucketType',
     'Cooldown',
     'CooldownMapping',
+    'MaxConcurrency',
 )
 
 class BucketType(Enum):
@@ -163,3 +167,129 @@ class CooldownMapping:
     def update_rate_limit(self, message, current=None):
         bucket = self.get_bucket(message, current)
         return bucket.update_rate_limit(current)
+
+class _Semaphore:
+    """This class is a version of a semaphore.
+
+    If you're wondering why asyncio.Semaphore isn't being used,
+    it's because it doesn't expose the internal value. This internal
+    value is necessary because I need to support both `wait=True` and
+    `wait=False`.
+
+    An asyncio.Queue could have been used to do this as well -- but it
+    not as inefficient since internally that uses two queues and is a bit
+    overkill for what is basically a counter.
+    """
+
+    __slots__ = ('value', 'loop', '_waiters')
+
+    def __init__(self, number):
+        self.value = number
+        self.loop = asyncio.get_event_loop()
+        self._waiters = deque()
+
+    def __repr__(self):
+        return '<_Semaphore value={0.value} waiters={1}>'.format(self, len(self._waiters))
+
+    def locked(self):
+        return self.value == 0
+
+    def wake_up(self):
+        while self._waiters:
+            future = self._waiters.popleft()
+            if not future.done():
+                future.set_result(None)
+                return
+
+    async def acquire(self, *, wait=False):
+        if not wait and self.value <= 0:
+            # signal that we're not acquiring
+            return False
+
+        while self.value <= 0:
+            future = self.loop.create_future()
+            self._waiters.append(future)
+            try:
+                await future
+            except:
+                future.cancel()
+                if self.value > 0 and not future.cancelled():
+                    self.wake_up()
+                raise
+
+        self.value -= 1
+        return True
+
+    def release(self):
+        self.value += 1
+        self.wake_up()
+
+class MaxConcurrency:
+    __slots__ = ('number', 'per', 'wait', '_mapping')
+
+    def __init__(self, number, *, per, wait):
+        self._mapping = {}
+        self.per = per
+        self.number = number
+        self.wait = wait
+
+        if number <= 0:
+            raise ValueError('max_concurrency \'number\' cannot be less than 1')
+
+        if not isinstance(per, BucketType):
+            raise TypeError('max_concurrency \'per\' must be of type BucketType not %r' % type(per))
+
+    def copy(self):
+        return self.__class__(self.number, per=self.per, wait=self.wait)
+
+    def __repr__(self):
+        return '<MaxConcurrency per={0.per!r} number={0.number} wait={0.wait}>'.format(self)
+
+    def get_bucket(self, message):
+        bucket_type = self.per
+        if bucket_type is BucketType.default:
+            return 'global'
+        elif bucket_type is BucketType.user:
+            return message.author.id
+        elif bucket_type is BucketType.guild:
+            return (message.guild or message.author).id
+        elif bucket_type is BucketType.channel:
+            return message.channel.id
+        elif bucket_type is BucketType.member:
+            return ((message.guild and message.guild.id), message.author.id)
+        elif bucket_type is BucketType.category:
+            return (message.channel.category or message.channel).id
+        elif bucket_type is BucketType.role:
+            # we return the channel id of a private-channel as there are only roles in guilds
+            # and that yields the same result as for a guild with only the @everyone role
+            # NOTE: PrivateChannel doesn't actually have an id attribute but we assume we are
+            # recieving a DMChannel or GroupChannel which inherit from PrivateChannel and do
+            return (message.channel if isinstance(message.channel, PrivateChannel) else message.author.top_role).id
+
+    async def acquire(self, message):
+        key = self.get_bucket(message)
+
+        try:
+            sem = self._mapping[key]
+        except KeyError:
+            self._mapping[key] = sem = _Semaphore(self.number)
+
+        acquired = await sem.acquire(wait=self.wait)
+        if not acquired:
+            raise MaxConcurrencyReached(self.number, self.per)
+
+    async def release(self, message):
+        # Technically there's no reason for this function to be async
+        # But it might be more useful in the future
+        key = self.get_bucket(message)
+
+        try:
+            sem = self._mapping[key]
+        except KeyError:
+            # ...? peculiar
+            return
+        else:
+            sem.release()
+
+        if sem.value >= self.number:
+            del self._mapping[key]
diff --git a/discord/ext/commands/core.py b/discord/ext/commands/core.py
index aa4343f5c..e587d3a72 100644
--- a/discord/ext/commands/core.py
+++ b/discord/ext/commands/core.py
@@ -33,7 +33,7 @@ import datetime
 import discord
 
 from .errors import *
-from .cooldowns import Cooldown, BucketType, CooldownMapping
+from .cooldowns import Cooldown, BucketType, CooldownMapping, MaxConcurrency
 from . import converter as converters
 from ._types import _BaseCommand
 from .cog import Cog
@@ -53,6 +53,7 @@ __all__ = (
     'bot_has_permissions',
     'bot_has_any_role',
     'cooldown',
+    'max_concurrency',
     'dm_only',
     'guild_only',
     'is_owner',
@@ -90,6 +91,9 @@ def hooked_wrapped_callback(command, ctx, coro):
             ctx.command_failed = True
             raise CommandInvokeError(exc) from exc
         finally:
+            if command._max_concurrency is not None:
+                await command._max_concurrency.release(ctx)
+
             await command.call_after_hooks(ctx)
         return ret
     return wrapped
@@ -248,6 +252,13 @@ class Command(_BaseCommand):
         finally:
             self._buckets = CooldownMapping(cooldown)
 
+        try:
+            max_concurrency = func.__commands_max_concurrency__
+        except AttributeError:
+            max_concurrency = kwargs.get('max_concurrency')
+        finally:
+            self._max_concurrency = max_concurrency
+
         self.ignore_extra = kwargs.get('ignore_extra', True)
         self.cooldown_after_parsing = kwargs.get('cooldown_after_parsing', False)
         self.cog = None
@@ -331,6 +342,9 @@ class Command(_BaseCommand):
             other.checks = self.checks.copy()
         if self._buckets.valid and not other._buckets.valid:
             other._buckets = self._buckets.copy()
+        if self._max_concurrency != other._max_concurrency:
+            other._max_concurrency = self._max_concurrency.copy()
+
         try:
             other.on_error = self.on_error
         except AttributeError:
@@ -718,6 +732,9 @@ class Command(_BaseCommand):
             self._prepare_cooldowns(ctx)
             await self._parse_arguments(ctx)
 
+        if self._max_concurrency is not None:
+            await self._max_concurrency.acquire(ctx)
+
         await self.call_before_hooks(ctx)
 
     def is_on_cooldown(self, ctx):
@@ -1800,3 +1817,36 @@ def cooldown(rate, per, type=BucketType.default):
             func.__commands_cooldown__ = Cooldown(rate, per, type)
         return func
     return decorator
+
+def max_concurrency(number, per=BucketType.default, *, wait=False):
+    """A decorator that adds a maximum concurrency to a :class:`.Command` or its subclasses.
+
+    This enables you to only allow a certain number of command invocations at the same time,
+    for example if a command takes too long or if only one user can use it at a time. This
+    differs from a cooldown in that there is no set waiting period or token bucket -- only
+    a set number of people can run the command.
+
+    .. versionadded:: 1.3.0
+
+    Parameters
+    -------------
+    number: :class:`int`
+        The maximum number of invocations of this command that can be running at the same time.
+    per: :class:`.BucketType`
+        The bucket that this concurrency is based on, e.g. ``BucketType.guild`` would allow
+        it to be used up to ``number`` times per guild.
+    wait: :class:`bool`
+        Whether the command should wait for the queue to be over. If this is set to ``False``
+        then instead of waiting until the command can run again, the command raises
+        :exc:`.MaxConcurrencyReached` to its error handler. If this is set to ``True``
+        then the command waits until it can be executed.
+    """
+
+    def decorator(func):
+        value = MaxConcurrency(number, per=per, wait=wait)
+        if isinstance(func, Command):
+            func._max_concurrency = value
+        else:
+            func.__commands_max_concurrency__ = value
+        return func
+    return decorator
diff --git a/discord/ext/commands/errors.py b/discord/ext/commands/errors.py
index 6087c1dfd..0d5e0d0f1 100644
--- a/discord/ext/commands/errors.py
+++ b/discord/ext/commands/errors.py
@@ -41,6 +41,7 @@ __all__ = (
     'TooManyArguments',
     'UserInputError',
     'CommandOnCooldown',
+    'MaxConcurrencyReached',
     'NotOwner',
     'MissingRole',
     'BotMissingRole',
@@ -240,6 +241,28 @@ class CommandOnCooldown(CommandError):
         self.retry_after = retry_after
         super().__init__('You are on cooldown. Try again in {:.2f}s'.format(retry_after))
 
+class MaxConcurrencyReached(CommandError):
+    """Exception raised when the command being invoked has reached its maximum concurrency.
+
+    This inherits from :exc:`CommandError`.
+
+    Attributes
+    ------------
+    number: :class:`int`
+        The maximum number of concurrent invokers allowed.
+    per: :class:`BucketType`
+        The bucket type passed to the :func:`.max_concurrency` decorator.
+    """
+
+    def __init__(self, number, per):
+        self.number = number
+        self.per = per
+        name = per.name
+        suffix = 'per %s' % name if per.name != 'default' else 'globally'
+        plural = '%s times %s' if number > 1 else '%s time %s'
+        fmt = plural % (number, suffix)
+        super().__init__('Too many people using this command. It can only be used {}.'.format(fmt))
+
 class MissingRole(CheckFailure):
     """Exception raised when the command invoker lacks a role to run a command.