Several years ago I ran Jetico I. It was in beta at the time and back then I did not understand why anyone would develop a firewall that responded to non network related events. Then came the leak testing craze and now many firewalls respond to non network events. I am not very enthusiastic about the way leak testing has become the dominant factor in firewall performance evaluation. None the less, Jetico I has decent, but not perfect leak test performance. One might observe that with LUA/SRP every leak test once downloaded by your browser will fail to run without elevating to administrative rights first. What about that zero day trojan game you just have to run? Well, you might have your firewall or HIPS turned off or install mode to install it. The reality is its not that hard to protect a computer against inadvertent program execution or drive by attacks, but there is really no automated way to protect the user against his own intentional actions. How about light? The question of a light firewall AV or suite comes up on a weekly basis around here. I don't know if all these people have P III's or just like snappy performance. Constantly people post screen captures of the task manager with the "mem usage" circled and they are completely missing the point. All that proves is good memory management, not lightness. On XP where you really need to look is on the performance tab, physical memory, available. And this has to be done on a before and after basis. A change that frees up 20 MB of mem usage might result in 40 MB of available physical memory. Matoused uses some test to measure the effect of a firewall on certain system tasks. Some of his published results ranged from a less than 20% slow down form ZA Pro to over 40% for Sunbelt. He must not be measuring the right thing because ZA Pro takes the cake for bloat with all sorts of non firewall features including a barely functional anti spyware scanner and record breaking mem usage. Ever enable the CPU Time column in task manager? You might start to throw out some software after you do. Comodo 2.4 uses very few CPU cycles until a P2P application is running, and then it goes nuts. Most people just look for CPU spikes, but this measures cumulative work of the CPU. A firewall that uses 1% on a cumulative basis may not seem like much, but it kills battery life on a notebook. ZA Pro uses at least 1% at idle, more with P2P running. Jetico I, Look 'N Stop and PC Tools use almost nothing at all. Its not just the line items in task manager for the firewall services you have to be looking at. Modern firewalls with HIPS change the CPU and memory utilization of other services as they run. Available physical memory magically shrinks. Seemingly unrelated services like touchpad drivers start to chew up CPU cycles in a big way. Sometimes it helps to give these trusted application status. Anyway, Jetico I passed every category with flying colors. I must be smarter now than I was then, because it has quieted down after a few hours, and I can see how to make it quieter if necessary. It runs in LUA without any problems. If you are on the creative side it has a lot of flexibility as it is rules based. If you must have the ultimate in leak proof living, add a hips or LUA/SRP. Remember security is a process, not a program.