1) Check the mating surfaces of the head and manifold for imperfections - if you can feel them, they're significant. Correct by draw filing (look it up if you need to - it's a useful thing to know...)
2) When you install the gasket next time, apply a thin (~1/16") even coat of RTV copper (nothing else - the red won't last...) to both sides of the gasket. Install gasket, install manifolds, install screws finger tight only. Go have a sammich and a cuppa - give it about a half-hour to start setting up. Maybe an hour. Come back, and torque to spec (start at top centre, clockwise spiral outwards, 19-21 pound-feet of torque.) Better to do to 10 pound-feet then to 19-21, since that will help the RTV "squidge" (technical term there!) into any remaning imperfections and such.
The RTV copper really is "copper-bearing silicone," and I use it on exhaust gaskets all the time. The Permatex stuff is also sensor-safe - I checked. The coating of RTV will give an extra boundary layer on both sides of the gasket, and will help to fill any imperfections that allow gasses to get past the "fire ring" (crimped metal ring around exhaust ports) that is likely causing your gasket to burn out.
Also:
Replace the screws and studs. I tend to use silicon bronze, but you can also replace with carbon steel if you can't find Si-bronze. Use SAE5 instead of SAE8 - and if you have to use carbon steel, replace them every time you do the job! They get weak with heat-cycling (although SAE5 holds up to repeated heat-cycling better than SAE8 does.) If you can get Si-bronze, then you can just anneal them and reuse them - heat them to a dull red (800* or so, I believe) and drop them in water. This reverses the "hardening" of the metal that takes place in use - and metal that is hardened without being tempered gets brittle (yes, I know it sounds backwards. Trust me - we're dealing with a copper-based alloy, and not an iron-based one. Copper responds to heating and quenching in the opposite manner as iron.)
If you put anything at all on the threads, use LocTite #272 or equivalent. No other grade of LocTite will survive the heat of "normal use," and you shouldn't need never-seez for this! (If by some chance you do decide to use never-seez, torque values are 5 pound-feet for intermediate and 10 pound-feet for final. But, you don't want that.) Why don't you want never-seez? Manifold screws work loose on their own, and don't need the help. If a manifold screw is loose enough long enough, you'll be doing the gasket job yet again! On that note, it's a good idea to check torque on those screws annually - you'll find one or two loose when you do. Don't skip the lower rear ones because they're a pain - those are usually the ones that are loose...
Make sure you retain and reuse the dish-shaped washers under the screw heads - they're there for a reason. That reason is to help keep the screws from loosening under normal use (iron expands .000006"/in/*F. Aluminum twice that. Screws that have to clamp aluminum and iron will have an uneven force under the head, that tends to unscrew the thing. The dish-shaped washers put a little "give" in the system to allow for this.) If you lose them, replace them - you can probably still get them OEM, or you can order Belleville Spring Washers from a fastener supply house. I'd probably use two under each screw head, nested together. I have yet to experiment with this idea, so I'm not sure.
If you need studs, you can cut them easily enough from threaded rod - 3/8"-16. Cut them about two inches long. Make sure to get new nuts - the OEM ones are threaded 3/8"-24 (there is a valid engineering reason for doing this sort of thing, but it doesn't apply here, as the loads are too small.) I use brass rod for new studs. Threaded rod in carbon steel is typically not treated to SAE8-equivalent strength, which is what you want. However, if you cut the studs from steel, coat the cut ends (since they won't be plated.)