You could try this out
select substr(a.tabschema,1,40) as TABLE_SCHEMA, substr(a.tabname,
1,40) as TABLE_NAME, card, stats_time, sum(avgcollen) as ROW_LENGTH
from syscat.tables a,syscat.columns b where a.tabschema = b.tabschema
and a.tabname = b.tabname and a.tabschema not like '%SYS%' group by
substr(a.tabschema,1,40),substr(a.tabname,1,40),card,stats_time order
by card desc with ur

On multiplying "card" and "ROW_LENGTH", we would get estimated size of
the tables in  bytes.

HTH
-Saurabh

Of course you can try this with the information taken from the
catalog. But the statistics have to be actual...

db2pd -db DB -tcbstats

gives a more reliable output - you can also specify the tablespace
(and table ?).

The size of the file is also a way - but this works for SMS only:
Derive the containers of the tablespace where your table lives.
Then the files have the tableid in their names in these containers.
There can be different types like *.dat  (data) *.INX *.IN1 (indexes)
and others.
They give you the current usage on disk for data and indexes and lobs.

db2pd also gives the sizes of indexes...

I found out that you could get the actual size of a table TABLE1 with this
SQL:

SELECT TABSCHEMA, TABNAME, DATA_OBJECT_P_SIZE + INDEX_OBJECT_P_SIZE +
LONG_OBJECT_P_SIZE + LOB_OBJECT_P_SIZE + XML_OBJECT_P_SIZE AS TOTAL_SIZE
FROM SYSIBMADM.ADMINTABINFO WHERE TABNAME = 'TABLE1'

-----------------------------~-~-~-~-~-~-~-~-~-~-

-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-