Exhaust The age old header question: How Big?


Active Member
Sep 3, 2018
Monrovia, California
More than any other question I get everyday, "How big should the primaries be on my long tube headers?" is still number one. More than "Are the primaries equal length?", or "Are the headers stainless steel or mild steel?", or even "How much are the Long Tube Headers?" It's a legitimate question, probably based on the theory that if the primaries are too small it will hurt power. I used the word "theory" because in many cases, smaller diameter primaries may boost power. There are many questions that need to be addressed before you decide on the diameter of the primary tubes on a set of headers, such as maximum expected horsepower, is it a race car, a street car, a street/strip car, engine displacement, transmission (stick or automatic), vehicle weight, cylinder head, and on and on. If the engine is a Small Block Ford (289, 302, 351W, etc.) there is typically one common issue, and it's the cylinder head and in particular, the exhaust valve. The stock exhaust valve on the Mustang 5.0L (302) engine is 1.45" in diameter. In the vast majority of aftermarket aluminum cylinder heads, the exhaust valve diameters have been increased to 1.60". This is a pretty big improvement (about 11%), but 1.60" diameter is still smaller than 1 5/8" (which is 1.625"). And we must remember that the exhaust does not flow through a 1.60" diameter hole (the outside diameter of the head of the valve) but through the valve seat, which is probably about 1 1/2" in diameter, which is further reduced by having the stem of the valve in the way. So, the real question should be, "Why would anyone want a primary tube larger in diameter than the initial restriction found around the exhaust valve area?"

Okay, I have not addressed the original question of "how big". If you are interested in this discussion, let me know and I will be more than happy to go into it further and provide some answers...

George Klass - RCI Headers
(909) 552-3690
[email protected]
Last edited:
  • Sponsors (?)

I love discussions about theory.. now there is no question that the valve is the smallest area, however the gasses are hotter in the cylinder than they are as they exit the exhaust. Is this more a question of velocity when dealing with primary sizes? conventional wisdom says larger tube is better.. but when you begin to discuss exhaust velocity vs temperature and the scavenging effect that has on the system I would think a more appropriately sized tube would make more power.
Good points, I was waiting for that. It's the discussion of a larger diameter tube moving more volume (exhaust gasses in this case) versus a smaller diameter tube that moves less volume, but moves the gasses at a higher velocity. Big tube equals more volume at a slower speed, small tube means less volume but at a higher velocity. On paper, it would look like the larger volume would make more power but the speed of the gasses leaving the exhaust port is important because of one issue, and that is "scavenging". As the exhaust gasses travel down the primary tubes, if they are moving fast enough, they can create a partial vacuum, which can then help pull the next charge with it. A large diameter tube, with it's slower moving exhaust gasses, may not have that advantage. Here is my rule of thumb. If you are running a light weight small block Ford powered drag race car, that leaves the starting line at or above 5,000 RPM, and you get your max power at or above 7,500 RPM, you will make more HP with a larger diameter primaries (1 7/8", 2", 2 1/8"), everything else being equal. On the other hand, if you are racing a small block Ford powered "street/strip" Mustang, that weighs 3,200 (or more) pounds, and have your maximum power at 6,000 to 6,500 RPM, you may get better results with 1 3/4" diameter primaries. With a "street/strip" Mustang at the drag strip, you need plenty of mid range power, from 3,800 RPM on up to your shift points. With the all out race car, that guy could care less about "mid range power", he's leaves the line at or above 5,000 RPM and is never below about 6,500 or 7,000 RPM going down the track. At this point, we have run head on into the most important rule about primary tube diameter, and it is the word "compromise". It has everything to do with what you are going to do with your Mustang. Is it mostly a race car or mostly a street car? The majority of our customers like to refer to their Mustangs as "street/strip" cars. I ask them, "How many miles do you put on your car in a normal year?". I would say that the average answer is between 8,000 and 10,000 miles a year (except in Los Angeles, where it's usually 15,000 miles). My next question is, "How many times do your race at the drag strip?" (street racing is meaningless for this discussion). The typical answer is that the average number of times they actually race at the track per year is about 5 to 6 times per year, and they average about four to six 1/4 mile passes. If they make six passes at the track, six times times per year, that is equal to 9 miles at the track, and 9,000 on the street. Folks, that is not a "street/strip" car, that is a street car that occasionally goes to the drag strip. Even if they put 90 miles on the drag strip, that is till only 1% of the total miles Your headers need to fit what you actually DO with your Mustang. RCI builds headers with your choice of primary sizes of 1 3/4", 1 7/8", 2", and 2 1/8" for Small Block Ford cylinder heads (and up to 2 1/4" for the Big Blocks). By the way, we have drag race customers that are making up to 900 HP with 1 3/4" primary headers...
  • Like
  • Useful
Reactions: 3 users
Everyone I know or ever heard of has only one head, the one they were born with. But if you were born as a Ford 302 (5.0L) or a 351W engine, you have many different heads to choose from, and pretty much any of them would be better than the one they stuck on you at the factory. Larger valves, better shaped combustion chambers and larger ports are just some of the advantages found in aftermarket Ford small block style cylinder heads, and you have a choice of cast iron or aluminum, too. The brand or type of cylinder head you choose may also affect your header selection. Let's start this discussion talking about the header flange bolt pattern.

This is an Edelbrock Victor Jr. cylinder head, a very good high performance street head. Any headers that you may already have that fit the stock head will bolt up to this style head. The stock bolt pattern is 2" center to center. Many (or most) of the aftermarket cylinder heads have this same bolt pattern. Because these are primarily high performance street heads, they were not designed for large diameter header primaries. Although we can make 1 7/8" headers fit on these type heads, it's not something we typically recommend. The larger size tubing has to be "dented" in on both sides near the flange to allow installation of the header bolts (and the 7/16th end wrench). In fact, all headers of any size are most likely going to need to be "dented" near the flange to allow for the bolts. Even the 1 1/2" stock factory installed shorty headers are dented in this location. The largest header we recommend for a head that only has the 2" bolt pattern is 1 3/4" primary header.

This is a World Products (sometimes called the Dart or Windsor Jr. head) cylinder head. There are several manufacturers that use this style flange bolt pattern. They have the standard 2" pattern and also an "optional" pattern. The optional pattern is usually a 3" center to center pattern, although we have seen some with 2 1/2" centers (Edelbrock Victor) and some with 2 3/4" centers (Ford N-351 and TFS-R heads) . This means that if you already have some headers that fit your stock head, they can also be bolted to this type of head, using the 2" centers. And if you wanted to use a 1 7/8" or a 2" primary header, you will be able to bolt to this style head, by using the optional wider bolt pattern, without the necessity of putting the "dent" in the tube near the flange. We have had customers ask if they could drill the flange on the headers they already have for the wider pattern, and the answer is yes, they can. The problem is, there would be no performance benefit, since their existing headers would still have the dent in the tubes where the tube is welded to the flange. The advantage of using the wider pattern is that the headers can be built without the dent in the tubes, eliminating any restriction in that area. RCI headers always use the optional bolt pattern that is on a customer's cylinder heads if possible, as it usually adds about 5 to 8 more horsepower.

And then there is this cylinder head, the Trick Flow Hi-Port heads. The exhaust ports on the stock Ford Small Block heads have always been one of their weak spots. The bend in the port between the valve seat and the port exit (where the header flange is) has always been too sharp, causing a restriction. In the 1980's, Trick Flow came out with a "raised port" head. Raising the port allowed the "bend" to be partially straightened out, thereby eliminating much of the exhaust flow restrictions. The hot Mustang race car crowd went for this head like a rat goes to cheese. I can remember when every hot "Pro 5.0" and "Street Outlaw" drag car had Trick Flow Hi-Port cylinder heads. The only problem with the heads "back in the day" was that they only had the stock 2" center to center bolt locations. RCI was the first company to design and offer "flange plates" for this head. A flange plate looks like a regular header flange, only it bolted to the heads using the 2" pattern, using counter sunk Allen bolts, and a wider (and threaded) diagonal pattern that would allow the racers to use 2" header primaries (the header flange bolted to the flange plates, that bolted to the engine). This was the only way for the racers to use a 2" primary header. At some point in the evolution of this head, Trick Flow began to include our (RCI's) diagonal pattern as an optional pattern for this head (which was fine with us). Now the big tube primary headers can bolt directly to the heads, eliminating the flange plates entirely. Other companies now offer "raised port" cylinder heads for those that are building all out small block Ford race engines, although not all the ports are raised the same distance. Fortunately, RCI has flanges for all the different cylinder heads, so it's no problem for us to offer a header that fits your specific heads.

To be in the header business, you also have to be able to manufacture flanges. There is no Ford type cylinder head that we don't have flanges for, and it's not just to fit the specific bolt pattern of the customer's cylinder head. Not all aftermarket small block Ford cylinder heads have the spark plugs in the same location, or that extend out at the exact same angle. The standard mass produced header companies offer headers designed to fit the OEM stock cylinder heads, it's basically designed to be a "one size fits all" header, which is why you hear about customers that use these mass produced type headers having to grind their flanges so they can get their spark plugs installed, or to clear enough room around the plugs to get the spark plug socket to go over the spark plugs. It's usually not an issue with stock type heads, but it can be a major pain in the butt with some of the higher performance aftermarket heads. All of our flanges are designed so that access to the spark plugs is a non-issue. Okay, enough of the "sales talk". If you need high performance headers for your Mustang (1965-1995), you know where to reach us...

George Klass - RCI Headers
(909) 552-3690
[email protected]
  • Like
Reactions: 1 user
Ok George, first let me tell you that I know your reputation for header building, you experience and knowledge on the subject is vast. That said, and keeping in mind that the majority of the members here have a street car, tell us how important is this to them. They are thinking 'all that for a losey 5 to 8 hp?' And how this relates to the average street driven stang.
I understand what you are talking about.
The 5 to 8 additional horsepower I was talking about is in regards to eliminating the "dents" next to the flanges by using a cylinder head with the optional bolt pattern. But as you say, if it truly is a street car, I doubt anyone would be able to feel much difference in the seat of their pants. On the other hand, there is no extra charge that RCI adds to the price of headers if they want them with flanges set up for the optional bolt patterns. It's basically a free 5 to 8 more HP. On a typical "hot" 5.0L Mustang, a set of 1 3/4" RCI Long Tube headers on an engine making between 350 and 375 hp will add about 20 to 25 more HP over the stock shorty headers, everything else being equal. Interestingly enough, the 1 3/4" RCI Mid Length headers are within 3 to 5 HP of the Long Tubes, throughout the entire RPM range. We have done numerous back to back dyno tests. The Mid Length headers use the same flanges and the same collectors as the Long Tubes. The only difference is the length of the primary tubes (and they cost less).
I've got a street car. Thats all its going to be. It currently has BBK long tube headers on it. They are rusted up and look terrible. At some point I'd like to change them out for something different. The long tubes are cool, but I was thinking about going back to shorty headers. Reading through your write up and looking at your site are the mid length headers a good compromise for a street car? Do they eliminate the headache associated with working with the long tubes or just move the issue to a new spot.

I'm not looking to do anything any time soon but curious to know the answer. If it makes a difference, SVO X302 heads, BBK SSI intake, no idea on cam, and honestly I want to change all that at some point as well.
I honestly feel that our Mid Length headers are a better choice for most street or even most street/strip cars. The horsepower difference is VERY small between the Long Tubes and the Mid Length headers, and they are much easier to install. And if you ever need to R&R your clutch or transmission, there is no need to remove the headers. They are available in primary tube sizes of 1 3/4", 1 7/8" and even in 2". We have all out drag only cars running the Mid Length headers. There is a guy in my neighborhood that is running a very serious bracket car. He has a 400+ inch 351-W, with a 'Glide trans in an '87 Mustang. He runs the 2" Mid Length headers, open, with no exhaust system or mufflers. He used to run our Long Tubes (2"), which he eventually damaged because of serious wheel stands. I convinced him to try the 2" Mid Lengths and he loves them. He changes converters alot at the tracks, and he said it's a breeze with the Mid Lengths, and they are lighter weight, too. He swears to me that he is running the same numbers as he did with our Long Tubes, and I believe him. He actually makes money with his car.
  • Like
Reactions: 1 user
Guys that have been drag racing as long as I have (my first run down the 1/4 mile was at the Santa Ana drag strip, in 1955, I was 16 years old) will tell you that so much of what goes on with hopping up your car is "monkey see, monkey do". Unless you are building a serious race only car, that is trailered to and from the track, and you are running in the 9's or quicker, I can usually guess that (A) your cam is too big and is probably hurting your mid range, (B) your carb (if you are not EFI) is too big and is hurting your mid range, and your header primaries are too big. Guys buy big camshafts, oversized carburetors and huge headers that are designed to work great at 7,800 RPM, and stick them on engines that rarely get past 6,500 RPM. The best header is the one that works with YOUR combination, not the combination of the guy down the street that likes to pop his hood as often as he can at the drive in and have the fans say, ooohh. I know this is a ford Mustang site, but I used to have a 409 Chevy when I was a pup (in 1962). A little street racing now and then, and a little Super Stock in between. I purchase the car new in Pasadena, from Service Chevrolet. The guy that ran the tune-up shop and ran the chassis dyno was Dyno Don Nicholson (I live just a few blocks from where he and his brother had their speed shop). He still had his '61 409 back then. That Chevy was cleaning up all over the nation in S/S. A local shop had built him a neat set of 1 3/4" headers. On a hunch, he built himself the same header only he did it in 1 5/8" tubing instead of the 1 3/4". The car picked up ET and 2 mph in the lights. How is that possible? His Chevy weighed 3400 lbs. and the engine would not make power above 6,000 RPM. The 1 5/8" primaries gave him additional torque and power in the mid range, where he needed it. Most Mustangs are like that, they weigh 3200 to 3400 lbs, and the 5.0L is usually out of steam much past 6,000 RPM. To be very honest, a lot of hot street driven Mustangs would run super with 1 5/8" headers, if they were GOOD headers and correctly designed (and that excludes every Shorty header and any header that uses a ball & socket type collector). But because of the Monkey See, Monkey do world that exists in our business, we could never make a living selling 1 5/8" headers, despite the reality that for many guys, that's actually what they need. Again, one more time. With all the aftermaket cylinder heads available, they all have a 1.60" diameter exhaust valve, and the valve seat where the exhaust has to run through is 1.50". It's like an hour-glass, where the sand on the top has to squeeze through the restricted area in the center, and winds up in the area on the bottom that is the same size as the area on the top.

Does anyone reading this think that by making the bottom half of this thing larger in diameter, that the sand is going to run past the restriction in the middle faster? Nope, it won't. The restriction is the restriction, and nothing you do underneath it is going to change the restriction. Exhaust gasses are not the same as sand, of course, we have temperature differentials going from very hot when it leaves the cylinder head to cooler as it goes down the pipe, but the principles are basically the same.
Gen. Karthief brings up an interesting subject, the word "torque", the measurement of a twisting motion. Torque and Horsepower are not two separate items. In fact, an engine dyno does NOT measure horsepower, it only measures torque. To find out the horsepower, the computer in the dyno has an algorithm formula that computes torque and RPM to come up with the horsepower reading. There is no way to increase horsepower without also increasing torque, or increasing torque without also increasing horsepower, but what you CAN do is to build your torque curve in an RPM range where you want your maximum horsepower.



I picked some generic engine dyno sheets off the internet. The line that starts at the lowest point on the graph is the HP line, and the other line is the torque reading. The HP and torque numbers are usually listed vertically on the graph, and the RPM numbers are usually horizontal. Let's just take the bottom graph for a minute. The maximum torque (light blue line) is listed at 221.5 lbs at about 6,300 RPM and the maximum HP (dark blue line) is listed at 308.6 HP at a little over 8,000 RPM. This is some type of a turbo Honda engine. The HP numbers go up in an almost straight line and the torque numbers are pretty consistent between about 5,300 RPM up to about 7,000 RPM. This would be called a "flat torque curve". Now, even though these three graphs are obviously three different types of engines with different torque and HP numbers, there is one thing on the dyno graphs that is exactly the same (and will be the same on every dyno graph you will ever see). Can you spot it?
I'm going to spare you the tension. On the graphs, the horsepower line crosses the torque line at some point, and that point is ALWAYS at 5,252 RPM. Remember when I mentioned that the dyno only measures torque and RPM, and the horsepower is computed by a formula or equation.

One horsepower is equal to 33,000 foot-pounds of work per minute. Add in the equations relating to torque and velocity, and you'll find that horsepower always equals torque multiplied by rpm, divided by 5,252. Canceling out the equal variables, you wind up with horsepower equaling torque... at 5,252 rpm, and once you have the HP at that number, you can extrapolate the HP at all the other RPM's. Sounds like a bunch of gobble-dee-gook, doesn't it, but that's the way it is.

Doesn't really help you to determine what size your header primaries should be though, does it?
Last edited:
Yes, we have done and continue to do dyno testing comparing primary tube sizes, but one fact continues to stand out. Large primary tubing may help HP in the upper RPM ranges, but it comes at the expense of mid range HP. On the other side, smaller primary tubing may help HP in the mid ranges but lose some at the higher RPM ranges. Of course, smaller and larger are relative numbers, how small is small and how large is large? Technically with small block Ford wedge heads, small is 1 5/8" and large is 2", a spread of 0.375" (a little more than 1/4" and less than 1/2"). It's even smaller if the header is a 1 3/4" (which is almost the defacto standard these days), in which case we are talking about a spread of 1/4". Like many things in life, there are compromises. The question is always "What am I willing to give up to have something else?" Again, the answer to that question is dependent on the answer to this question, "What am I going to do with this car?" If it's going to be a dual purpose car, which of the two purposes are going to be dominant over the other? If most of your driving is going to be on the street, and keeping the engine under 6,200 RPM, with only an occasional blast at the drag strip, that is a different header than one on the same car that is a race car, with an engine that peaks HP above 7,000 RPM and that is almost never driven on the street. I have to assume that this is obvious to everyone. And yet, too many enthusiasts seem to think "big" instead of "the right" size, and that also includes the race car crowd too. I have a customer that has a '90 Mustang, 351W, pretty tricked out engine, and it runs pretty strong (it's a street/strip car but mostly strip). He told me that the car weighs about 3,000 lbs. with him in it. He had been running a 5-speed and recently swapped it for a Powerglide. He had been running a 2" header with the 3 1/2" collectors (standard on the 2" headers). The speed and ET had dropped off a little after the trans swap, said it just didn't seem to pull as hard as it did before. We talked for awhile and I reminded him that he went from a 5-speed to a 2-speed transmission. The Powerglide is a great transmission, and is very consistent (my customer is a bracket racer), but it requires a little more mid range power. With the 5-speed he was using, he was shifting at 7,300 RPM which dropped down to about 6,700 RPM on each shift. This meant that he was operating in a power band between 6,700 and 7,300 RPM. The 'Glide has only two speeds, "low" and "drive". When he shifted at 7,300 RPM the RPM dropped down considerably more. The wider your power band, the more the header size needs to be dealt with. A typical street car needs to have a power band somewhere between 3,000 and 6,250 RPM, give or take. My customer allowed me to talk him into going with a 1 7/8" header (with the standard 3" collector) to pick up a little in the mid range and he said he can feel it pulling hard again. He lost about 1 MPH over the 5-speed but the ET came back to what he was running before. Another example of "everything is dependent on everything else". Next season we are going to try a 1 7/8" stepped to a 2" header (with a 3 1/2" collector), just to see if we can push the MPH up a little bit. We have been building step headers for years, like the 1 3/4" stepped up to a 1 7/8" and the 1 7/8" stepped up to a 2". I'm not really that pumped up about them myself, sometimes they help and sometimes you can't see much if any difference. On a street car, I have never seen an advantage, on a street/strip or a race car, sometimes I do and sometimes I don't. It's all just like the secret of life, "what is, is, and what ain't, ain't"...
  • Useful
  • Like
Reactions: 1 users