If you're talking about sorting based on, across all variables, the total number of missing values in each row, try using the CMISS function. It's new to SAS 9.2, and if you supply all the columns in your data set as arguments, it will tell you how many of them have missing values. It's similar to the NMISS function, but CMISS can take arguments of character or numeric variables (or any combination!). Whatever number CMISS returns is what you'd sort on.
FYI, PROC SQL and hence the query builder do not support variable lists. So, sadly, you can't use the syntax "cmiss(of _all_)" in the Query builder expression editor. You could write your own DATA step code where that would be allowed.