The OP wrote me offline and indicated:

"I want the term date to be . (as it shows up on the duplicate Row 3)... so if there is a duplicate - grab EFFDate from the first one (earliest) and Termdate from the second one (latest or missing)."

Unless someone has a better idea, I would just do it in a data step using first. and last. E.g.,

data have; infile datalines dsd missover; input ID effdate termdate; informat effdate mmddyy10.; informat termdate mmddyy10.; format effdate termdate date9.; datalines; 1,10/14/2009,. 6,08/01/2008,01/02/2012 6,10/14/2009,. ;

proc sort data=have; by descending id effdate; run;

data want (drop=first_effdate); set have (rename=(effdate=first_effdate)); retain effdate; format effdate date9.; by descending id; if first.id then effdate=first_effdate; if last.id then output; run;

HTH, Art --------- On Thu, 3 Dec 2009 07:46:45 -0800, Sdlentertd <sdlentertd@GMAIL.COM> wrote:

> have this dataset >ID EFFDATE TERMDATE >1212 10/14/2009 . >625 08/01/2008 10/14/2009 >625 10/14/2009 . > >I need to remove duplicates but leave the duplicate's Original >EFFDATE >there. The dataset will look like this: > > >ID EFFDATE TERMDATE >1212 10/14/2009 . >625 08/01/2008 . > > >manupulating duplicate's dates Only. Duplicate is based on ID field >only. P.S. Sorting a and resorting does NOT work, it leaves the old >date in, but I need for EFFDate to be taken from one duplicate reccord >and TERMDATE to be taken from the other duplicate record. >Thank you