"Accurate" is relative.
Dynos are a necessary evil. To a builder, they're a way of making sure the engine doesn't come apart, it doesn't leak, doesn't smoke, carb is tuned, timing is right, etc. It's also a great way to show incremental changes. To the general public, it's a bragging right, and most guys just wanna see that peak hp number.
This is something I discuss with almost every other builder, including Barry, Keith, and Blair, because it's a "thorn in the side" to all of us.
From what I've seen in the use of several engine dynos in the area, most do not read the same. This was one of the factors behind the EMC...you get a bunch of guys on the same dyno, then all the guessing and variables can go away. I'm sure Barry or Blair will tell you the same thing, but as a builder, you basically pick a dyno that you're familiar with and then stick with that one so you can have a way of comparing past/future builds. Barry has his own. Blair uses Jim Morgan's dyno. I use Dale Meers' dyno. That way, if I do a 482 with a specific cam/head/intake package, then change a cam on the next 482 while keeping the other variables the same, I can see what the difference is.
I'm often concerned that the numbers I'm getting are "accurate". The correction factor has a large part in the numbers and where the weather station is located has a large part in the numbers. If the computer is getting STP from the intake air, it will show different numbers than if the computer is getting STP from the dyno room, where the engine is running and the headers are adding to ambient temperature.
On the dyno I use, it's typical to see correction factors anywhere from 1.06 to 1.1. That's a 6-11% correction right off the bat, and it will vary based on whether I dyno in the winter, dyno in the summer, etc. I've even taken the time to gather weather data independently and calculate the correction factor manually, just to compare to what the dyno is seeing. Again, the general jist of it is that if I dyno that same engine in the winter or summer, the numbers should be the same because the computer should be correcting.
I have already made plans with Barry to dyno one of my engines here, then make some pulls up there. I'm curious to see the difference as we often compare notes and commonly see the same horsepower between similar builds, but I tend to show more torque. If I dyno here and then dyno there and show a lot less torque, then it will be more data for us to sift through and decipher. To be brutally honest, we all want to know how we compare to each other. I'll be the first one to raise my hand and say that I'm not as sharp as Blair, but it would be nice to get us all on the same dyno one day to see what the difference actually is.