Well I think there is a important difference between the UK and unique index, however this difference comes into play only with 8i and above with the emerge of "deffered constraints". Preveously UK and PK could not have been implemented without the corresponding underlying *unique* index, so there realy wasn't any difference. But now you can implement UK and PK on *nonunique* index, so that you can temporrarilly deffer constraint.

So with unique index your data must allways be unique, while unique constraint allows you to temporarilly have nonunique values in a column, which sometimes comes very handy.

BTW, what sonaliak mentioned about the requirement that all unique_key columns must be referenced in a where clause for the index to be used is certanly not correct.

HTH,