As I mentioned before, I had my problems with amulegui. Sometimes it worked, sometimes it didn't - with the same binary.
There have been issues with the memory management, as it ate up all free memory. After that, it either just disappeared or stayed and did business as usual.
That's the reason, why I am suspicious about the loading of the server list. This is the part of amulegui, which is displayed, when starting. And the crash appears after showing "Loading..." (bottom, left side) for about 40 seconds, so my guess is, that sth in EC is responsible.
On my linux box I can't see anything out of the order. The log tells me sth about an external client
2005-12-27 09:15:08: Neue externe Verbindung akzeptiert
2005-12-27 09:15:08: Verbinde zu Client: amule-remote 0x0001
2005-12-27 09:15:08: Zugang gewährt.
2005-12-27 09:15:48: Externe Verbindung getrennt.
The client has the same reproducible crash at the same address (varying with each new build, of course) , reading the same memory location 0x34 (which in fact is just an offset, the real address being 0).
I'll try to get GNU Debugger for MingW, so that I can produce one of your much loved backtraces

The other possibility would be, if one of you developers could point me to some code parts, where to implement some debug prints (or send me a patch), which can be helpful to corner the evil bug, before shooting it. Would it help to have a debug build of amuled running on my linux box?
Meanwhile I'll fetch the new snapshots and try if sth has changed.