A deadlock occurs when two or more
processes try to access a resource that another process has
a lock on. Microsoft SQL Server automatically
detects these deadlocks, terminates one of the processes,
and rolls back any changes made by that process.
The process chosen for termination will be the one with the
least amount of active time on the server. In
most cases, you can simply reissue the statements that were
rolled back and should have no problems.

Deadlocks occour when two or more processes place a lock on 
the same resources and each process waits for the others to 
release the lock.
The options to avoid deadlocks are basically the following:
1) always update data in the same order: if process A 
updates table t1 and then table t2 and process B updates 
table t2 and then table t1 deadlocks can occour. Choose an 
order and use it everywhere in your code.
2) issue the commands again when you catch the deadlock
3) always update data before data selection: no other 
process will place a lock on the modified records
4) identify the processes that can run with lower deadlock 
priority and issue SET DEADLOCK_PRIORITY LOW before these 
statements: this will tell SQLServer that the process is a 
good candidate for killing in case of deadlock, having 
the "important" processes free to run without kills

Hi 
1) 
First find out which process block the other process from 
activity monitor or sp_who2. Then kill that process. 
2)
If update locks are more  then it will make deadlock. So 
reduce that.