I've been a supporter of 3.90.3 for a long long time, but wanted to convert some of my new cds using lame version 3.97b1 (and from today, also b2 - same results).

This mostly is classical music, piano concerts, opera stuff.

Using the V2 vbr-new paramater, bitrates for my new cds were 122-160kpbs.
I found this scarily low, although mind you, they sound great.

So I used the V2 without vbr-new (and reading the forums, the target bitrate still is the same), and these same cds are now 160-200kpbs. A significant difference, I'd say!
And roughly the same as 3.90 would give.

I thought it might be contributable to the 128kbps bottom no longer existing, but the number of samples that are below that are very limited, 128-160 is mostly chosen, whereas the non-vbr-new tends to choose 160-224.

Is there an explanation for this enormous difference in bitrate between V2 and V2 --vbr-new?

up. I have wondered this as well.

This is one of the improvements in --vbr-new. Don´t panic. The developers try to tune the psy-model for not wasting bits while even getting better overall quality. There is much tuning going on therefore.
Maybe it isn´t that accurate on every piece of music atm but that is why it´s called beta.
Please report any degraded sample you hear harmed by --vbr-new...

Edit: I want to add that my statement above shouldn´t give the wrong impression. Not that accurate on every piece of music means that there are rare exceptions.