When performing the following query as either a SP or from Query 
Analyser,there is a substantial difference in performance between using 
variables for the date range (>50sec) and hard coding the date (instant). 
The tbl_unit_result_log.date field has an index. The index does not appear 
to be used when variables are implemented (Table Scan), but are, when hard 
coding the dates.- This table has 1.8 million rows.


declare @user_tag nchar(5)
declare @startd datetime
declare @endd datetime

set @user_tag='2001'
set @startd='2001-10-16 0:0:0'
set @endd='2001-10-16 23:0:0'

SELECT
    tbl_Unit_Result_Matrix.description,
    count(tbl_Unit_Result_Log.Result_Type)
FROM
     tbl_Unit_Result_Log INNER JOIN tbl_Admin_Users ON
        tbl_Unit_Result_Log.User_Tag = tbl_Admin_Users.User_Tag
     INNER JOIN tbl_Unit_Result_Matrix ON
        tbl_Unit_Result_Log.Result_Type=tbl_Unit_Result_Matrix.Result_Type
WHERE
    tbl_Unit_Result_Log.Date  between @startd and @endd
    and tbl_Unit_Result_Log.User_Tag=@user_tag
    group by tbl_Unit_Result_Matrix.description
 
Does anyone know how the performance can be improved when using variables

Thank you