I believe you cannot do it with distinct, but you may bypass it as in the example below,

data t1;
   do i=1 to 20;
      a=ceil(ranuni(127)*5);
   output;
   end;
   keep a;
run;
proc print;run;
proc sql;
create table t2 as
  select a as b, min(N) as ord
  from (
         select a , monotonic() as N
         from t1
        )
  group by 1
  order by 2
  ;
  quit;
  proc print;run;

原表添加自增ID列
按此列排序

能具体些么？谢谢

他的意思是
data dataset1;
set dataset1;
id=_n_;
run;

proc sql;
create table dataset2  as
select distinct a as b
from dataset1
order by id;
quit;

我试了一下，不行的，你没必要按照a的原来顺序排列，因为a里面有些重复出现的不一定以次出现。一般用sql distinct的话，系统会自动排序，如果是以次出现的话可以用下面的程序去做：
data a;
input a;
cards;
3
3
1
1
2
;
run;
data b;
set a;
d=lag(a);
retain n 1;
if _n_=1 then n=1;
else do;
if a^=d then n+1;
end;
run;
proc sql;
create table c  as
select distinct n,a as b
from b
order by n;
quit;