In the long run you might want to restructure your data to have mulitple observations instead of repeating all of those columns. Otherwise this can be done with array, especially now that SAS allows array references in PUT statements. array applicanttype_ (3) ; array type_ (3) ; array status_ (3) ; array code_ (3); array indicator_ (3) ; array fil_ (3); do i=1 to dim(fil_);   if applicanttype_(i) ne ' ' then put       "PES" (type_(i) status_(i) code_(i)) ($3.)             indicator_(i) $1.  fil_(i) $3.   ; end; ... View more

A little pointer control and you can do it pretty simply: data foo; infile cards missover; input @'PERS:PERS1|' pers1    $  8.        @'ADD:P1A1|'   add1_1   $ 20.        @'ADD:P1A2|'   add1_2   $ 20.        @1        @'PERS:PERS2|' pers2    $  8.        @'ADD:P2A1|'   add2_1   $ 20.        @'ADD:P2A2|'   add2_2   $ 20. ; cards; PERS:PERS1|MIGS    ADD:P1A1|5 HIGH ST           PERS:PERS1|ROBBIE  PERS:PERS2|KYLIE   ADD:P1A1|ROSE COTTAGE        ADD:P1A2|ROSE COTTAGE        ADD:P2A1|SOMEWHERE           ADD:P2A2|SOMEWHERE           ; run; ... View more

Hi Haikuo, catx(' ','nmiss(',name,')as ' ) will create a string "nmiss(variable-name)". nmiss(variable-name) will create a number. maybe you treat "'nmiss(',name,') as' " "'nmiss(',name,') as'" It should be "'nmiss(',name,') as'" ... View more

Building on Tom's suggestion, if the set of random names is in a dataset called otherNames then you could use : proc sql noprint; select name into :randNames SEPARATED BY ' ' from otherNames; select count(name) into :nameCount from otherNames; data want ; set have; name = scan("&randNames.",ceil(&nameCount.*ranuni(0))); run; PG ... View more