The query has to be fully optimizable.

CLEAR
CREATE TABLE mytable (fld1 I, fld2 C(20))
INSERT INTO mytable VALUES (1, "A")
INSERT INTO mytable VALUES (1, "B")
INSERT INTO mytable VALUES (1, "C")
INSERT INTO mytable VALUES (2, "AA")
INSERT INTO mytable VALUES (2, "BB")

INDEX ON fld1 TAG fld1
INDEX ON DELETED() TAG deleted
SELECT * FROM mytable WHERE fld1 = 1 INTO CURSOR crsResult
? DBF(), "-", FILTER()

>I remember an issue where reccount() could be misleading after a query is not using nofilter. Anyone else remember that or have an example. Or am I just wrong.