Query Optimisation on SE 5.0

I have a system under SCO Unix 4.2 with SE 5.0 engine and RDS 4.10 which is doing a large (12 table) join. The problem is that the query optimiser is taking a perverse route slowing the query down substantially and this is happening because it thinks sequential access is *much* cheaper than indexed access - costs given by SET EXPLAIN are around 200 to 17000 if I query this table separately. Because this table is about 4 levels down in the join hierarchy it will only extract around 2000 from 26000 records but the solution proposed by the optimiser is to pivot the whole query around sequentially scanning this table and joining outwards in two directions. Any ideas? - apart from Online or hacking the data in systables! Paul Hatcher Graduate Associates Ltd +44 81 840 3110 - Phone phatcher@cix.compulink.co.uk +44 81 567 1894 - Fax