Casting vs using the ‘as’ keyword in the CLR

When programming interfaces, I’ve found I’m doing a lot of casting or object type conversion.

Is there a difference between these two methods of conversion? If so, is there a cost difference or how does this affect my program?

public interface IMyInterface
{
    void AMethod();
}

public class MyClass : IMyInterface
{
    public void AMethod()
    {
       //Do work
    }

    // Other helper methods....
}

public class Implementation
{
    IMyInterface _MyObj;
    MyClass _myCls1;
    MyClass _myCls2;

    public Implementation()
    {
        _MyObj = new MyClass();

        // What is the difference here:
        _myCls1 = (MyClass)_MyObj;
        _myCls2 = (_MyObj as MyClass);
    }
}

Also, what is “in general” the preferred method?