[
lruskauff, no I hadn't yet specifically reproduced the OPs problem, only that I could reproduce the above problem with items that were and were not within the only a single category. Other than that, when picking two items there are 16 different attribute combinations to potentially test assuming that grows for any second product tested all assuming that the individual wasn't logged in when the issue is possible to be duplicated/observed in a linear checkout process... Each new variable to possibly try after this point simply multiplies the number of variations needed to try to replicate a problem that has previously been observed if no further assistive information is provided... I might get lucky, but I doubt it myself at "finding" the way to reproduce the OP problem through complete random testing as at the moment this would involve me making a complete purchase at the store through PayPal purchasing at least 2 items per purchase to eventually see in a purchase that I only got charged for one of them...
Ok, I guess I'm not following the Paypal part so I'll go reread. Are you saying tho that it only will show after we send it to Paypal? I was hoping it wasn't that, makes it hard to check. But then again, didn't the OP replicate it, then clear the cart but couldn't replicate it again. So if he had to clear the cart, he had not gone to a payment module, so why would we?

Where did he go by the way?