Or do you always want 1 in case you have 1 and anything else in case you have only one record?

If so, my original query would be modified as

row_number() over (partition by PrimaryKey order by case when ChildTableField = 1 when -100000000000 else ChildTableField end) as RowNum

to make sure 1 comes first.

>I have a SQL which returns some records. The result may look like this:
>
>

>PrimaryKey PrimaryKeyChildTable ChildTableField
>1                    1                  0
>4                    5                  1
>8                    6                  2
>

>
>However, in some particular situations, depending on this big SQL syntax, we might end up with more than one records from the master table such as:
>
>

>PrimaryKey PrimaryKeyChildTable ChildTableField
>1                    1                  0
>4                    5                  1
>4                    6                  2
>5                    8                  1
>5                    8                  2
>8                    6                  2
>

>
>So, on top of the SQL syntax I have presently, I would like to add the necessary code so I would filter any record from a duplicate primary key at the child table level and only keep the record from the child table having ChildTableField=1. So, in such circumstances, when I would have two records from the child table in the query result, I would like to obtain this:
>
>

>PrimaryKey PrimaryKeyChildTable ChildTableField
>1                    1                  0
>4                    5                  1
>5                    8                  1
>8                    6                  2
>