data temp1;

input x y;
datalines;
1 3
4 5
;
data temp2;

input x y;
datalines;
2 6
3 7
;
run;
data temp;
set temp1(rename=(x=x1 y=y1));
set temp2(rename=(x=x2 y=y2)) ;
x=x1+x2;
y=y1+y2;
run;

data temp1;

input x y;
datalines;
1 3
4 5
;
data temp2;

input x y;
datalines;
2 6
3 7
;
run;
data temp3;
input x y;
datalines;
3 4
5 6
;
run;
data temp4;
input x y;
datalines;
4 5
6 7
;
run;
data temp5;
input x y;
datalines;
5 7
8 9
;
run;
%macro temp(num);
data temp;
%do i=1 %to &num.;
set temp&i.(rename=(x=x&i. y=y&i.));
%end;
x=sum(of x1-x&num.);
y=sum(of y1-y&num.);
drop x1-x&num. y1-y&num.;
run;
%mend temp;
%temp(5);

还有个办法是直接转成矩阵，利用矩阵相加运算，再转回来。

楼上的你写一个IML程序给我们看一下
我想学习一下 谢谢

1# 甜甜居士
最方便的方法就是用矩阵计算PROC IML

proc iml;
use temp1;read all into temp1;/*读入数据集temp1 temp2*/
use temp2;read all into temp2;
close temp1;close temp2;
total=temp1+temp2; /*计算temp1 temp2 每个元素的和*/
create total from total;
append from total; /*将结果导出到total数据集*/
quit;