Hello, i've a small question. The following code gives me an merge join cartesian *********** ** ** **** ** select distinct vzd.nr_rel_vzd nr_rel_vzd , vzd.nr_bron nr_bron from (select nvl(max(ins_nr),0) max_ins_nr from eur_fct_nat_pers where id_bron = (select max(nr_bron) from eur_el_vzd where rownum = 1)) fct , eur_el_vzd vzd , eur_el_rel_adr_click rac where vzd.ind_fict = 'n' and vzd.nr_rel_vzd = rac.nr_rel(+) and vzd.nr_rol_vzd = rac.nr_rol_rel(+) and vzd.nr_bron = rac.nr_bron_rel(+) and (vzd.upd_nr >= fct.max_ins_nr or vzd.ins_nr >= fct.max_ins_nr or rac.upd_nr >= fct.max_ins_nr or rac.ins_nr >= fct.max_ins_nr) *********** ** ** **** ** explain plan select statement, goal = choose cost=10985 cardinality=53808 bytes=3067056 sort unique cost=10985 cardinality=53808 bytes=3067056 filter hash join outer merge join cartesian cost=2086 cardinality=124750 bytes=4491000 view object owner=dv_soft cost=1056 cardinality=1 bytes=13 sort aggregate cardinality=1 bytes=8 partition range single partition hash all table access full object owner=mart object name=eur_fct_nat_pers cost=1056 cardinality=133955 bytes=1071640 sort aggregate cardinality=1 bytes=4 count stopkey table access full object owner=dv_el object name=eur_el_vzd cost=1030 cardinality=249500 bytes=998000 table access full object owner=dv_el object name=eur_el_vzd cost=1030 cardinality=124750 bytes=2869250 table access full object owner=dv_el object name=eur_el_rel_adr_click cost=667 cardinality=242117 bytes=5084457 *** ** **** *** *** *** this is because the where conditions has an or statement in it. I can't get this merge join cartesian out. I've tried a union (and loose the or statement) but this is not good for the performance. I've already joint all tabels in the were clause. Anyone got an good suggestion? regards, richard

Kindly post your question clearly.