From previous talks this semester we have a good idea of how Martin-Löf random density-one points behave. For example, they form a proper subclass of the difference random reals (the so-called "Madison randoms"), and are hence incomplete. We investigate their behavior in the absence of randomness. We will show that the notions of dyadic density-one and full density-one are not the same (even though they coincide on the randoms). We will also show that every real is computable from a full density-one point.