>>> from decimal import Decimal
>>> dec = Decimal(2 ** 1024)
>>> dec
Decimal('179769313486231590772930519078902473361797697894230657273430081157732675805500963132708477322407536021120113879871393357658789768814416622492847430639474124377767893424865485276302219601246094119453082952085005768838150682342462881473913110540827237163350510684586298239947245938479716304835356329624224137216')
>>> dec += Decimal('0.1')
>>> dec
Decimal('1.797693134862315907729305191E+308')
>>> dec == Decimal(2 ** 1024) + Decimal('0.1')
True
>>> dec -= Decimal(2 ** 1024)
>>> dec
Decimal('2.109752663820230210576934273E+280')
>>> dec == Decimal('0.1')
False
>>>

You don't say what behaviour you were expecting!   :)

By design, almost all Decimal operations (but not creation of a Decimal from an integer or string) round to the precision given by the current context.  By default that precision is 28 significant digits.  See the documentation, particularly the quick-start tutorial,

http://docs.python.org/library/decimal.html#quick-start-tutorial

for more information, including how to change the default precision.

It might also help to note that, with decimal, what you see (from repr() of a Decimal instance) is *exactly* what you get.  So when you see

>>> dec
Decimal('1.797693134862315907729305191E+308')

that number really is exactly what's stored in dec:  there are no 'hidden'  digits waiting to appear.