> Can it be depended on the type of query or Server maybe?
>
> Did a " select count(*) from table where InvoicePrinted = 'T' "
>
> InvoicePrinted is a Char;
>
> No index took 3 secs, with an index on InvoicePrinted took 0.1 secs

Even a index with very low selectivity is good for selects.
With a multi-million-rows table, a index for a column with maybe
10 unique values could save your hours of retrival time.

Indexes with very low selectivity are mainly a problem, when you
insert or update data. This results in old row-versions, which must
be garbage collected. And a sweep on a such a database could
easily take days.

HTH,
Tord