filter — Selective Aggregates

The filter clause extends aggregate functions (sum, avg, count, …) by an additional where clause. The result of the aggregate is built from only the rows that satisfy the additional where clause too.


The filter clause follows an aggregate function:

SUM(<expression>) FILTER(WHERE <condition>)

With the exception of subqueries and window functions, the <condition> may contain any expression that is allowed in regular where clauses.0

The filter clause works for any aggregate function: besides the well-known functions such as sum and count, it also works for array_agg and ordered set functions (e.g., listagg).

If an aggregate function is used as a window function (over clause), the syntactic order is: aggregate function, filter clause, over clause:

SUM(...) FILTER(WHERE ...) OVER (...)

However, the filter clause is not generally allowed before over rather, it is only allowed after an aggregate function, but not after other window functions—it is not allowed with ranking functions (rank, dense_rank, etc.) for example.

Use Cases

The following articles describe common use cases of filter:


SQL:2003 introduced the filter clause as part of the optional feature “Advanced OLAP operations” (T612). It is barely supported today, but is easy to emulate using case (see Conforming Alternatives).

Conforming Alternatives

Generally, the filter clause can be implemented using a case expression inside the aggregate function: the filter condition has to be put into the when-clause, the value to be aggregated into the then clause. Because aggregate functions generally skip over null values,1 the implicit else null clause is enough to ignore non-matching rows. The following two expressions are equivalent:

SUM(<expression>) FILTER(WHERE <condition>)
SUM(CASE WHEN <condition> THEN <expression> END)

Count(*) needs some special treatment because “*” cannot be put into the then clause. Instead, it is enough to use a non-null constant value. This ensures that every matching row is counted. The implicit else null clause maps non-matching rows to null, which is ignored by count too.

COUNT(*) FILTER (WHERE <condition>)

When using a set quantifier (distinct or all) it must remain in the aggregate function prior the case expression.

On my Own Behalf

I make my living from training, other SQL related services and selling my book. Learn more at

Proprietary Extensions

PostgreSQL: Subqueries Allowed

The PostgreSQL database supports subqueries inside the filter clause (e.g., via exists).

Proprietary Alternatives

As the above described alternative with the case expression is very widely supported I recommend using that approach rather than the proprietary alternatives offered by some products.

Countif (Google BigQuery)

Google BigQuery offers the aggregate function countif similar to Microsoft Excel.

COUNTIF(<condition>) [OVER(…)]

The same result can be obtained with the case approach, which is very widely supported.

COUNT(CASE WHEN <condition> THEN 1 END) [OVER(…)]

About the Author

Photo of Markus Winand

Markus Winand is the SQL Renaissance Ambassador. He is on a mission to introduce developers to the evolution of SQL in the 21st century. Markus can be hired as trainer, speaker and consultant via

Buy his Book on Amazon

Cover of “SQL Performance Explained”: Squirrel running on grass

The essence of SQL tuning in 200 pages

Buy on Amazon
(paperback only)

Paperback and PDF also available at Markus’ store.

Hire Markus

Markus offers SQL training and consulting for developers working at companies of any size.
Learn more »



SQL:2016-2: §10.9, Syntax Rule 8.

When emulating using case only window functions are disallowed. SQL:2016-2: §10.9, Syntax Rule 7a.


SQL:2016-2: §10.9, General Rule 7a (for regular aggregates), 8a (for binary set operations), 10d (for inverse set operations). See also SQL:2016-2: §4.16.4.

Exception: array_agg (SQL:2016-2: §10.9, General Rule 12gii, explicitly in note 510) — the subquery array construction can remove null values without filter clause.

Connect with Markus Winand

Markus Winand on LinkedInMarkus Winand on XINGMarkus Winand on Twitter
“modern SQL” by Markus Winand is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 Unported License.
Legal | Contact | NO WARRANTY | Trademarks | Privacy and GDPR | CC-BY-NC-ND 3.0 license