BK

Was the original router also bridging over the Frame Relay link? I would agree with Paolo that a routed implementation is more normal and would be easier to troubleshoot. But if it was working ok before on the old router this way then it should be able to work that way with the new router.

If it is passing Internet traffic then we can probably assume that it is not a basic connectivity or addressing problem. Have you been able to determine if the traffic that is failing is sent over the Frame Relay and the head end is not routing it properly or if the remote router is not sent over the Frame Relay?

It might be helpful if you would post the configuration of the BVI and of the Frame Relay class referenced on the subinterface.

I note that there is an ip helper-address configured on the Frame Relay serial interface. Since this interface has no IP address the helper-address will not function (because IP is not enabled on this interface). The helper-address should be configured on the BVI interface.

HTH

Rick

The problem is that going from a release to another certain things can change subtly and stop working, just because cisco has decided that something wasn't right before, at the cost of breaking it after.

Being the application bridging, we aren't helped here by the well know diagnosing techniques but have to use something else, be it deduction, trial-and-error, or mac address scrutiny.

This is a why a sane design has more chances of surviving technology changes.