Why not use T-SQL to generate the queries to be executed?  Then cut and paste the result of this into a text file to be executed using OSQL.

For example,

SELECT DISTINCT
'SET NOCOUNT ON SELECT C. CustomerID, C. CompanyName, C.City, O. OrderDate, O. Freight FROM Customers C INNER JOIN Orders  O  ON C.CustomerID = O.CustomerID WHERE C.CustomerID = '''+C.CustomerId+''''
FROM Customers C
     INNER JOIN
     Orders O ON C.CustomerID = O.CustomerID

Since folks are casting around for the "more than one way to do it" - my tendency would be to use perl and sqsh.   Has the advantage of working equally well from a Unix or Windows client, and works whether the database is MS SQL Server, Sybase, or both.

Which simply reflects my own background and circumstances as does the original authors.

Cannot be used in the situation we are discussing at present. This T-SQL would generate statements for every single customer that made an order, but what we need is something to handle a large subset of a large customer base provided in the form of text file.

In other circumstances T-SQL would do just great. I use it often especially for admin tasks.

At my company we run into "requests" like this all the time.  Us lowly developers don't have the right to create jack in the production database but the data results the business unit wants come from there.  Our normal solution is to use Excel.  In a Query analyzer window we'll do a create table statement to create a temporary table to hold the data, in this case "create table #tmp (cust varchar(5))"

We'll have a list of clients sent to us in Excel or a text file.  Opening the attachment in Excel.  We'll insert a blank column in front of the data column and define a formula as ="insert into #tmp values('" & A2 & "')"  Copy the formula in A1 to the entire A column.  Copy the column and do a Past...Special...Values.  Copy the column of insert statements into QA and run.  Now you have a temp table in a QA session you can join to the data in the database and use in whatever ways one needs.

We deal with payroll data for millions of people employed throughout the country and it is not uncommon to get a request to pull demographic info on 10-20K people at a time for some report or another.  This method makes these requests trivial to handle.

SET NOCOUNT ON

SELECT
    C.CustomerID,
    C.CompanyName,
    C.City,
    O.OrderDate,
    O.Freight
FROM
    dbo.Customers AS C
    INNER JOIN dbo.Orders AS O ON C.CustomerID = O.CustomerID
    INNER JOIN OPENDATASOURCE(
        'Microsoft.Jet.OLEDB.4.0',
        'Data Source=c:\;Extended Properties="text;HDR=No;FMT=Delimited"'
    )...custID#txt AS T ON T.F1 = C.CustomerID