JOIN Efficiency Using Multiple ON Conditions Versus WHERE Conditions

My question is fairly simple. When I join between two tables, I always use the ON syntax. For example:


SELECT

*
FROM

Users

JOIN UserRoles

ON (Users.UserRoleId = UserRoles.UserRoleId)


No problems there. However, if I then decide to further filter the selection based on some trait of the UserRole, I have two options: I can add the condition as a WHERE statement, or I can add the condition within the ON block.

--Version 1:

SELECT

*
FROM

Users

JOIN UserRoles

ON (Users.UserRoleId = UserRoles.UserRoleId)
WHERE

UserRoles.Active = 'TRUE'


-- Version 2

SELECT

*
FROM

Users

JOIN UserRoles

ON (Users.UserRoleId = UserRoles.UserRoleId

AND UserRoles.Active = 'TRUE')


So, the question is, which is faster/better, if either? The Query Analyzer shows the two queries have the exact same execution plan, which makes sense, since they're both joining the same tables. However, I'm wondering if adding the condition in the ON statement results in fewer rows the JOIN statement initially needs to join up, thus reducing the overall initial size of the results table before the WHERE conditions are applied.

So is there a difference, performance wise? I imagine that if Users had a thousand records, and UserRoles had 10 records, then the JOIN would create a cartesian product of the two tables, resulting in 10,000 records in the table before the WHERE conditions are applied. However, if only three of the UserRoles is set to Active, would that mean that the resulting table, before applying WHERE conditions, would only contain 3000 records?

Thanks for whatever information you can provide.