Do a left join to the parent table on the key in question, then examine the values in the child table where the value in the left-joined parent table are null.
For example, if this was your schema...
table1:
myKey int primary key,
...other columns...
table2:
otherKey int primary key,
myKeyFromTable1 int
...other columns...
You'd do this:
select distinct
t2.myKeyFromTable1
from table2 t2
left join table1 t1 on t1.myKey = t2.myKeyFromTable1
where t1.myKey is null
That would give you the distinct values in table2
that wouldn't have a corresponding parent in table1
.