To sort a QUERY by more than one column, list the columns after order by, separated by commas, each with its own direction:
=QUERY(A1:F8, "select A, B, D order by A asc, D desc", 1)
That sorts by column A from A to Z and, inside each group of equal values in A, by column D from largest to smallest. This guide shows how that works on real data, how to sort by totals, and the three things that most often spoil a sorted QUERY: blank rows, the header row and a missing comma.
Tested in Google Sheets. On November 11, 2026 we ran every query on this page against the same seven-row sales table, including deliberately broken versions, and copied the result or the exact error message. How we test
Want the formula without writing it? Our QUERY formula generator builds the sort, the filters and the totals for you.
The sample data
| A: Region | B: Rep | C: Date | D: Amount | E: Status | |
|---|---|---|---|---|---|
| 2 | North | Ana | 2026-01-15 | 120 | Paid |
| 3 | South | Ben | 2026-02-03 | 80 | Open |
| 4 | North | Cai | 2026-01-20 | 250 | Paid |
| 5 | East | Ana | 2026-03-01 | 80 | Paid |
| 6 | South | O’Neil | 2026-01-05 | 40 | Open |
| 7 | North | Ben | 2026-02-28 | 100 | |
| 8 | East | Cai | 2026-02-10 | 250 | Paid |
Two columns, two directions
order by A asc, D desc returned:
| Region | Rep | Amount |
|---|---|---|
| East | Cai | 250 |
| East | Ana | 80 |
| North | Cai | 250 |
| North | Ana | 120 |
| North | Ben | 100 |
| South | Ben | 80 |
| South | O’Neil | 40 |
Column F held a score that none of these queries use. The second column only matters when the first one ties. Swap them and the meaning changes: order by D desc, B asc put both 250 rows first, then 120 and 100, then the two 80 rows with Ana before Ben, then 40.
ascis the default:order by A, D descgave exactly the same result asorder by A asc, D desc.- The keywords ignore case:
DESCworked likedesc. - Text sorts alphabetically, numbers by value and dates from oldest to newest.
order by C desclisted March 1 first.
Sort by a column you don’t show
The sort column doesn’t have to be in select. select B order by D desc returned only the names, ordered by amount: Cai, Cai, Ana, Ben, Ben, Ana, O’Neil.
Sort by a total
With group by, sort by the aggregate exactly as you wrote it in select:
=QUERY(A1:F8, "select A, sum(D) group by A order by sum(D) desc", 1)
It returned North 470, East 330, South 120. The heading of the total column was “sum Amount”; add label sum(D) 'Total' at the end to rename it. That gave the same rows with the heading “Total”.
Every column in select must be grouped or inside an aggregate. select A, B, sum(D) group by A failed with ADD_COL_TO_GROUP_BY_OR_AGG: B.
Top N rows
Add limit after the sort: select B, D order by D desc limit 3 returned Cai 250, Cai 250 and Ana 120. In the query text, limit goes after order by.
Problem 1: blank rows at the top
Open-ended ranges such as A1:C are handy because new rows are included automatically. But the empty rows below your data are included too, and in an ascending sort they come first. On a five-row table in another tab, select A, C order by C over A:C returned 14 empty rows before the first item.
Filter them out on a column that is always filled:
=QUERY('Week 8 data'!A:C, "select A, C where A is not null order by C", 1)
That returned just the five items, from 5 to 50. Empty cells in a sorted column behave the same way: order by E put the row with no status first.
Problem 2: the header row sorted into the data
Found in testing: with 0 as the third argument, QUERY treats the first row as data. =QUERY(A1:F8, "select A, B order by A", 0) sorted the header “Region, Rep” in among the rows, between North and South. Use 1 when your range starts with a header row. If you leave the argument out, Google’s help page says QUERY guesses; writing 1 removes the guesswork.
Problem 3: syntax errors
| Mistake | What we got |
|---|---|
Missing comma: order by A desc D |
PARSE_ERROR: Encountered ” <ID> “D “” at line 1, column 29. Was expecting one of: … “,” |
A column that isn’t in the data: order by Z |
NO_COLUMN: Z |
Lowercase letters: select a, d |
NO_COLUMN: a |
The column number in a PARSE_ERROR counts characters in the query text, so it points at the exact spot.
When SORT is simpler
If you don’t need to filter or total, the SORT function does the same with column numbers instead of letters. =SORT(A2:E8, 1, TRUE, 4, FALSE) (column 1 ascending, then column 4 descending) returned the same order as our first QUERY. SORT takes the data without the header row.
Filtering before you sort
A where clause goes before order by. Quick rules, all tested on the same data:
- Text in single quotes:
where E = 'Paid'. Comparisons are case-sensitive:where A = 'north'found nothing, whilewhere lower(A) = 'north'found the three North rows. - Dates with the word date:
where C > date '2026-02-01'. Without it, we got no rows and no error. - Several conditions with
andoror:where D >= 100 and E = 'Paid'. - A value from a cell is joined from outside the quotes:
"select B, D where D > "&G1.
The QUERY formula generator page lists the other mistakes we ran into.