On 01/16/2005 05:15:54 PM, Alfonso Le?n wrote: > Hello: >=20 > What is the recomended aproach to collect the statistics? >=20 > Thanks in advance for your suggestions My advice is to collect statistics in two passes: 1) Estimate (compute) statistics for tables and indexes. 2) Collect column histograms Two passes are necessary, because collecting column histograms misses =20 row statistics for the table. Here is how a typical "COMPUTE" script =20 would look like, in my world: DBMS_STATS.GATHER_DATABASE_STATS ( estimate_percent =3D> NULL, block_sample =3D> FALSE, method_opt =3D> 'FOR TABLE', granularity =3D> 'DEFAULT', cascade =3D> TRUE, options =3D> 'GATHER'); DBMS_STATS.GATHER_DATABASE_STATS ( estimate_percent =3D> NULL, block_sample =3D> FALSE, method_opt =3D> 'FOR COLUMNS SIZE SKEWONLY', degree =3D> NULL, granularity =3D> 'DEFAULT', cascade =3D> FALSE, options =3D> 'GATHER'); --=20 Mladen Gogala Oracle DBA -- //www.freelists.org/webpage/oracle-l