Yes, it is possible by using in memory table variables on 
SQL server 2000 and above.

A table variable can be created and all disticnt values can 
be inserted into this table and then deleting all rows from 
first table you can insert back disticnt rows to back to 
original table. Table variable would automatically be 
destroyed whenever bacth execution finishes.

using cursor is another option

sno         names 
1           arun
2           arun
3           arun
6           arun
7           bala
11          bala
12          guna
9           guna
14          guna
10          raj
13          raj

Table Name T1
Fields Names Sno,Names
Delete From t1 where sno not in 
(select min(sno) from t1 group by names)

Jaare Kameeney

select * into #empnew from (select * from emp union select 
* from emp) a

yes its very much possible.
create table #student
(name varchar(10),age integer)

insert into #student values ('A',12)
go 2
insert into #student values ('B',13)
go 3
insert into #student values ('C',15)
go 4
insert into #student values ('D',14)
go 5


SET ROWCOUNT 1
DELETE #student FROM #student A WHERE (SELECT COUNT(*) FROM 
#student B WHERE A.NAME = B.NAME AND A.AGE = B.AGE) > 1
WHILE @@ROWCOUNT > 0
DELETE #student FROM #student A WHERE (SELECT COUNT(*) FROM 
#student B WHERE A.NAME = B.NAME AND A.AGE = B.AGE) > 1
SET ROWCOUNT 0