>>re: 1. If the ID doesn't exist you will get an empty table. I'd worry about it at the client end.
>>
>>re: 2. Unless your table record is extraordinarily large, pulling * makes maintenance easier assuming ID is a primary or unique key.
>
>It is very wide table.

Are you pulling this record across the internet or a WAN, or just a LAN?

Unless you have varbinary or text type fields, even a record with a LOT of fields still should move across even a slow network almost immediately.

The safest bet for performance (and security if HIPPA data) is to only move the fields you really need. The safest bet for maintenance is to pull *.

I don't think I've run into a situation where pulling a single record from a table made a performance impact even across an internet connection (SQL server to Web server).

YMMV