Filter realtime Tweets

Overview contents

Please note:

We launched a new version of the POST statuses/filter endpoint as part of Twitter API v2: Early Access. If you are currently using any of these endpoints, you can use our migration materials to start working with the newer endpoint.



Returns public statuses that match one or more filter predicates. Multiple parameters may be specified which allows most clients to use a single connection to the Streaming API. Both GET and POST requests are supported, but GET requests with too many parameters may cause the request to be rejected for excessive URL length. Use a POST request to avoid long URLs.

The track, follow, and locations fields should be considered to be combined with an OR operator. track=foo&follow=1234 returns Tweets matching “foo” OR created by user 1234.

The default access level allows up to 400 track keywords, 5,000 follow userids and 25 0.1-360 degree location boxes. If you need elevated access to the Streaming API, please use the enterprise streaming API, PowerTrack.

Streaming message types

Standard 'Track' stream messages

In addition to standard Tweet payloads, the following kinds of messages may be delivered on a stream. Note that this list may not be comprehensive—additional objects may be introduced into streams in the future. Ensure that your parser is tolerant of unexpected message formats.

When parsing Tweets, keep in mind that Retweets are streamed as a status with another status nested inside it. If you are matching Tweet fields using regular expressions, it is possible that you will match fields in the nested Tweet instead of the wrapper. As a rule of thumb it is better to use a JSON parser to extract information from message payloads.


Maintenance Messages

Blank lines

On slow streams, some messages may be blank lines (\r\n or similar) which serve as “keep-alive” signals to prevent clients and other network infrastructure from assuming the stream has stalled and closing the connection.

Limit notices (limit)

These messages indicate that a filtered stream has matched more Tweets than its current rate limit allows to be delivered. Limit notices contain a total count of the number of undelivered Tweets since the connection was opened, making them useful for tracking counts of track terms, for example. Note that the counts do not specify which filter predicates undelivered messages matched.


Disconnect messages (disconnect)

Streams may be shut down for a variety of reasons. The streaming API will attempt to deliver a message indicating why a stream was closed. Note that if the disconnect was due to network issues or a client reading too slowly, it is possible that this message will not be received.

"code": 4,

The following table lists possible status codes and their meanings. Additional codes may be used without warning.

Code Name Description
1 Shutdown The feed was shutdown (possibly a machine restart)
2 Duplicate stream The same endpoint was connected too many times.
4 Stall The client was reading too slowly and was disconnected by the server.
5 Normal The client appeared to have initiated a disconnect.
7 Admin logout The same credentials were used to connect a new stream and the oldest was disconnected.
9 Max message limit The stream connected with a negative count parameter and was disconnected after all backfill was delivered.
10 Stream exception An internal issue disconnected the stream.
11 Broker stall An internal issue disconnected the stream.
12 Shed load The host the stream was connected to became overloaded and streams were disconnected to balance load. Reconnect as usual.

Stall warnings (warning)

When connected to a stream using the stall_warnings parameter, you may receive status notices indicating the current health of the connection. See the stall_warnings documentation for more information.

"message":"Your connection is falling behind and messages are being queued for delivery to you. Your queue is now over 60% full. You will be disconnected when the queue is full.",
"percent_full": 60


Compliance Messages

Status deletion notices (delete)

These messages indicate that a given Tweet has been deleted. Client code must honor these messages by clearing the referenced Tweet from memory and any storage or archive, even in the rare case where a deletion message arrives earlier in the stream that the Tweet it references.


Location deletion notices (scrub_geo)

These messages indicate that geolocated data must be stripped from a range of Tweets. Clients must honor these messages by deleting geocoded data from Tweets which fall before the given status ID and belong to the specified user. These messages may also arrive before a Tweet which falls into the specified range, although this is rare.


Withheld content notices (status_withheld, user_withheld)

These messages indicate that either the indicated Tweet or indicated user has had their content withheld.


These events contain an id field indicating the status ID, a user_id indicating the user, and a collection of withheld_in_countries uppercase two-letter country codes. This example illustrates a hypothetical Tweet that has been withheld in Germany and Argentina.

"withheld_in_countries":["DE", "AR"]


These events contain an id field indicating the user ID and a collection of withheld_in_countries uppercase two-letter country codes. This example illustrates a hypothetical user who has been withheld in Germany and Argentina.



User update

Everytime a user updates their profile we broadcast a user_update event to indicate that an update has been made to the user profile. The source and target objects are identical in content.

"created_at": "Tue Aug 06 02:23:21 +0000 2013",
"source": {
"target": {
"event": "user_update"

Was this document helpful?

Thank you

Thank you for the feedback. We’re really glad we could help!

Thank you for the feedback. How could we improve this document?

Thank you for the feedback. Your comments will help us improve our documents in the future.