SQL substring group by

Here's one for ya. I'm baffled at this.

this query returns 2455 results

SELECT * FROM (
      SELECT	SUBSTRING(ORDERNO, 1, 6) AS ORDERNO,
		        SUM(DOLLARS) AS DOLLARS
      FROM TEMPORDR 
     GROUP BY ORDERNO 
  ) TBL1
WHERE DOLLARS >= 500.00
ORDER BY DOLLARS

Select allOpen in new window

and this query returns 2792 results:

SELECT * FROM (
SELECT	SUBSTRING(ORDERNO, 1, 6) AS ORDERNO
		,SUM(DOLLARS) AS DOLLARS
FROM TEMPORDR 
GROUP BY SUBSTRING(ORDERNO, 1, 6)
) TBL1
WHERE DOLLARS >= 500.00
ORDER BY DOLLARS

Select allOpen in new window

the order number look like this 368954*1, 368954*2, etc. The *1 denotes each level on an order so I want to ignore the suffix and group the order number part of the order number. i.e. 368954.

When I use the substring is should return less results because it should group the like terms. However it returns a lot more. Anyone know why?