>>Maybe it wasn't a good idea to start this thread usings ints in the example. AFAICS the real problem is that PIVOT needs an aggregating function and you can't do that with a uniqueidentifier.....
>
>Here is dynamic Pivot using your example:
>
>

>declare @Cols varchar(max), @SQL nvarchar(max), @ColsWithNames varchar(max)
>
>select @Cols = coalesce(@Cols + ',','') + QUOTENAME(ColumnID), 
>@ColsWithNames = coalesce(@ColsWithNames + ', ','') + QUOTENAME(ColumnID) + ' as Column' + CAST(Row as varchar(10)) 
> from (select CAST(ColumnID as varchar(36)) as ColumnID, 
>ROW_NUMBER() over (order by  CAST(ColumnID as varchar(36))) as Row from test50 group by CAST(ColumnID as varchar(36))) X
>
>--print @Cols
>
>set @SQL = 'select RowID, ' + @ColsWithNames + 
>' from (select cast(ID as varchar(36)) as ID, cast(ColumnID as varchar(36)) as ColumnID, RowID from test50) src 
>PIVOT (min(ID) FOR ColumnID IN (' + @Cols + ')) pvt'
>
>--print @SQL
>
>execute (@SQL)

Thank you! Had to tweak it a little to get it to work with my tables but it works great. I'm still pawing through it to understand the pivot bit though.....
As a final twist: the data I really want to put into the new table won't be the Id field but another (varbinary) field from the source table. I haven't yet worked out whether the above can be modified to do that ?