Café de Nuit

The DMG (3.0, it may have been corrected in 3.5) doesn't ever state how you calculate CRs for monsters you've created. Is there any definitive way to calculate CRs besides playtesting that you've come up with?

Not definitive, no.

Playtesting is the only one. Any other way of calculating CRs proves faulty in one way or other. With all the flexibility offered by DnD you can never tell the CR of a new creature without involving human judgement at some point. The trick is not in what abilities the monster possesses, but how well those abilities perform when put together.

I agree with Igor. While I have a system to calculate CR, it is always a matter of judgement.
While the calculation may result in "CR 4" The result may be 3 or 5... Or sometimes 6.

And as my players always comment, I tend to overestimate a character's power (or a player's creativity) and underestimate a creature.

Since I mostly create monsters instead of picking them out (except when I don't have time, like in random encounters) I can tell you that not even playtesting is always correct.

What it may be a CR 7 for a certain party, it may be CR 5 for another party. Example:

Flesh Golem vs: 2 wizards, a cleric and a rogue (7th level) and the same flesh Golem vs 2 fighters, a ranger (Archer) and a cleric.
Cleric casts magic weapon on whatever weapon isn't magical (or Bull's strength or prayer etc) as the golem approaches (can't run, while the characters can, so they easily get 1 -2 rounds to repare) and then the fighters hack it appart easily. The DR 5/adamantine isn't that much)

In the October 2000 (I think) issue of Dragon Magazine, there was an article called 'How to Create a Monster'. At the end it had a CR "guesstimater" and I have found this article very useful for creating monsters and calculating CRs fairly accurately, but it still helps to playtest.

Of course...EL is a whole other can of worms.

I tend to handle experience that way.

Is the monster getting the jump on the party etc etc...