I'm wondering if anyone can suggest a trouble-free firewall for me to use?
I just seem to have problems with any I try. First I tried Firestarter which I found very easy to use and was showing all ports stealthed on every firewall test I tried. However, it seemed to cause problems with my dial-up modem, as soon as I installed firestarter the modem wouldn't work and every time I tried to connect it would freeze my comp.
After re-installing the drivers it would work again for a while then it would stop working again.
Next, I tried the Firewall-Knoppix that came with Kanotix and that would show all ports stealthed except icmp which was shown as open on every test no matter what I tried.
Now, I'm trying GuardDog which again is easy to use and with this one everything I have set to allow connects but web pages are loading really slowly.
What I really want is a rule-based firewall as much like Kerio 2 as possible, which I use on windows, and preferably with a GUI.
I was forever switching between firewalls on Windows til I found Kerio so I will probably be doing the same on Linux now til I get one that feels right.
Any recommendations.

one of the best gui firewall apps for linux is probably fwbuilder has just about every options under the sun interface it more like checkpoint

i personally like shorewall its been very nice i've tried quite a few firewall frontends myself but shorewall seems to be the perfect one for me but its only command line with no gui.

Well I'm still using Guarddog at the moment and it's working very well for me now, I've done a few firewall tests and all ports are stealthed. I think my problems were caused by the fact I'm so used to using Kerio on Windows that I found Guarddog a bit too confusing but I've got the hang of it now.
However, FWBuilder looks promising so I may give that a go.
I'm just wondering how good Linux firewalls are in general at stopping trojan hijacked programs from accessing the net? Do they use md5 or anything like that?

since i've never used guarddog i cant say but as far as trojans you will have to create rules for your firewall within guarddog to drop packets that go to certain trojan ports like if destination port x is used for a trojan then it would be best to drop that port. I doubt that it has any built in rules for these but i could be wrong

As far as md5 im not sure what you are talking about. There is hardly anyway for a firewall to compute a md5 on a packet and match that with something since it would be so much overhead and packets timestamp would be different hence a different md5. You could probably do it if you had and IDS or IPS which would reject certain kind of trojans and attacks based on rules something like snort which you would want to work in conjunction with your firewall.