At the risk of being redundant, repeating myself, and saying the same things over and over again, you might want to take a look at the following
lecture notes from MIT. I think lecture 16 has most of what's pertinent. Note also that the drag equations in Rocksim are rule-of-thumb approximations. To do better than rule-of-thumb you need to solve the equations of fluid dynamics...tightly coupled PDE's. Apogee offers AeroCFD that solves a reduced form of the Navier-Stokes equations, making steady-state ideal fluid assumptions to make the equations more *tractable* so that they can quickly solve the resulting potential flow equations with a panel method.
Bottom line is this: to state that Rocksim is *accurate* or *not accurate* is grossly oversimplifying things. You start here: what questions do you want to answer with Rocksim??? Do you want to predict altitude with absolute precision? Are you willing to conduct wind tunnel tests of your model to precisely determine Cd throughout the flight envelope and weigh all engines so that you will fly only with those that are identical to Rocksim specs? If so, then there is some chance that Rocksim will do what you want.
Personally, I don't fly that way. My models are built by hand, not by machines with low tolerances. I grab an engine out of my field box, ram it in my model, and fly it. It goes up and comes down. I repeat this several times and that is my day flying rockets. Never once do I recall saying to myself, "by gum, Rocksim said my rocket would fly to 732.6 feet and it did!"
I use Rocksim to answer the following questions:
1) Is my design likely to be statically stable?
2) What engines/delays are good for my model?
3) What recovery device is appropriate for my model for the field and conditions I'm flying in?
4) If it is a competition model, which design will fly higher or stay up longer?
For the first question Rocksim is reasonably accurate, but I still take precautions for that first flight.
Rocksim is accurate at helping me choose engines and delays.
Rocksim is sufficiently accurate to help me choose the right recovery device, although a specific flight may experience wind gusts or thermals that carry my rocket off the field.
Rocksim is generally sufficiently accurate to capture the trends that will help me choose between one competition design or another, but I don't trust it enough to let it have the final say.
All in all, Rocksim is accurate enough to give me the answers to my questions. Whether it is accurate enough for your purposes depends on what questions you are asking. The NARAM R&D report that's been discussed will be extremely valuable if it helps us determine which questions Rocksim can answer and to what degree of precision.