Take the following Dataset Merge example:

Data xx ; id=1; input x @@ ; datalines ;
1 2 3
;
run ;

Data yy ; id=1 ; x = 9; y=9 ; output; run;

proc sort data=xx; by id ; run ;
proc sort data=yy; by id; run;

Data merged; merge xx yy ; by id; run;
proc print; run;

What happens is this:

1. The first observation on dataset XX is read in, and the value of X is 1.
2. The first observation on dataset YY is read in.  The value of X is 9, and it overlays the value read in from dataset XX.
3. The merged observation is written out.
4. The second observation from dataset XX is read in, and the value of X is 2.
5. Since the ID values still match, no attempt is made to read another observation from dataset YY, the value of X does not change, and the merged result is written out.
6. Same thing happens with the third observation from dataset XX.
   

Many thanks to the contributors on the SAS Google Group for this lesson.