How do I generate random number for each row in a TSQL Select?

Take a look at SQL Server - Set based random numbers which has a very detailed explanation.

To summarize, the following code generates a random number between 0 and 13 inclusive with a uniform distribution:

ABS(CHECKSUM(NewId())) % 14

To change your range, just change the number at the end of the expression. Be extra careful if you need a range that includes both positive and negative numbers. If you do it wrong, it's possible to double-count the number 0.

A small warning for the math nuts in the room: there is a very slight bias in this code. CHECKSUM() results in numbers that are uniform across the entire range of the sql Int datatype, or at least as near so as my (the editor) testing can show. However, there will be some bias when CHECKSUM() produces a number at the very top end of that range. Any time you get a number between the maximum possible integer and the last exact multiple of the size of your desired range (14 in this case) before that maximum integer, those results are favored over the remaining portion of your range that cannot be produced from that last multiple of 14.

As an example, imagine the entire range of the Int type is only 19. 19 is the largest possible integer you can hold. When CHECKSUM() results in 14-19, these correspond to results 0-5. Those numbers would be heavily favored over 6-13, because CHECKSUM() is twice as likely to generate them. It's easier to demonstrate this visually. Below is the entire possible set of results for our imaginary integer range:

Checksum Integer: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Range Result:     0 1 2 3 4 5 6 7 8 9 10 11 12 13  0  1  2  3  4  5

You can see here that there are more chances to produce some numbers than others: bias. Thankfully, the actual range of the Int type is much larger... so much so that in most cases the bias is nearly undetectable. However, it is something to be aware of if you ever find yourself doing this for serious security code.


When called multiple times in a single batch, rand() returns the same number.

I'd suggest using convert(varbinary,newid()) as the seed argument:

SELECT table_name, 1.0 + floor(14 * RAND(convert(varbinary, newid()))) magic_number 
FROM information_schema.tables

newid() is guaranteed to return a different value each time it's called, even within the same batch, so using it as a seed will prompt rand() to give a different value each time.

Edited to get a random whole number from 1 to 14.


RAND(CHECKSUM(NEWID()))

The above will generate a (pseudo-) random number between 0 and 1, exclusive. If used in a select, because the seed value changes for each row, it will generate a new random number for each row (it is not guaranteed to generate a unique number per row however).

Example when combined with an upper limit of 10 (produces numbers 1 - 10):

CAST(RAND(CHECKSUM(NEWID())) * 10 as INT) + 1

Transact-SQL Documentation:

  1. CAST(): https://docs.microsoft.com/en-us/sql/t-sql/functions/cast-and-convert-transact-sql
  2. RAND(): http://msdn.microsoft.com/en-us/library/ms177610.aspx
  3. CHECKSUM(): http://msdn.microsoft.com/en-us/library/ms189788.aspx
  4. NEWID(): https://docs.microsoft.com/en-us/sql/t-sql/functions/newid-transact-sql