Hi Markus,
>OTOH this is only about 0.1 seconds faster than my original version but way worse to read. Part two of the contest (remove duplicates and create a table with a different structure) is still more than 17 seconds. The commands I used are INDEX ON...UNIQUE, COPY TO and SELECT...INTO TABLE.

Continuing on my assumption on eliminating the write to table1a:

CREATE TABLE Table2 ;
		(Name	C(61), Strasse C(34), Ort C(36), eMail C(30) )
FLOCK("Table2")
SELECT Table1
SCAN
	INSERT INTO Table2 VALUES ;
		(		RTRIM( Table1.Vorname ) +" "+ Table1.Name;
			,	RTRIM( Table1.Strasse ) +" "+ Table1.HausNr ;
			,	Table1.PLZ + Table1.Ort, Table1.eMail)

        *** too bad that the CSV has to be built AFTER Table2 is finished <bg>.
endscan

gives more work to the interpreting engine of vfp, but wins because of less disk activity. AND ETecs compiler should help making the scan loop less of a hit since it is JITed via IL.

regards

thomas