We've all heard the talk. It seems to run the gamut from, "Dude--headers are worth 100 hp," to the naysayers who snivel "Headers? They aren't worth the trouble." The truth is actually somewhere in the middle. Because results can vary so wildly, we decided to take a look at some past testing just to see what works and what doesn't.
If you're new to performance engines, and all headers tend to look alike to you, you might want to start with Jim McFarland's treatise on header design on page 62. This will give you the basics on why headers are designed the way they are, and why they don't all look the same. For this story, we're going to perform several header comparisons to shed some light on that age-old question of how much power headers are really worth.
All the dyno tests in this section were performed on different small-block Chevys, but when it comes to headers and exhaust tuning, this same information is completely relevant to any engine--only the header sizes change to match the displacements. This is a relatively busy subject with plenty of variables, but once you get to know all the important characters in this horsepower and torque stage play, the ending is always fairly predictable. Most of all, have fun with this. This is one session involving lots of hot air that's actually helpful.
 | 
Cast-iron manifolds are designed to merely connect the exhaust port with the exhaust pipes. Headers use mandrel-bent tubing to improve exhaust flow and reduce restrictions. Even on a stone-stock 350ci small-block Chevy, a simple set of 15/8-inch headers are worth as much as 53 lb-ft of torque and 34 hp. | 
Direct replacements like these HPC-coated DynoMax headers for an '88-'95 Chevy pickup are relatively short compared to long-tube headers, but they still offer a performance advantage over stock exhaust manifolds. |
The question of equal-length headers continues to be argued. Stahl and Headers by Ed both sell true equal-length headers that offer the potential for additional power, especially in the midrange rpm band. True equal-length headers are also much more expensive than production-style headers. | 
Tri-Y headers connect two pairs of pipes from each bank into two small Y-pipes and then join those larger pipes into a small collector. These photos show how the driver-side small-block header connects cylinders No. 1 and No. 5 into one pair and No. 3 and No. 7 into the second pair. This also steps the pipe diameters from four 1 5/8-inch primaries to a pair of 1 7/8-inch pipes. | 
Flange thickness does nothing to help performance, but when it comes time for your next set of headers, consider that thicker flanges that measure 0.300 inch or thicker are more stable with a superior sealing surface that will prevent header leaks and blown gaskets. |
Large-tube headers merely move the torque peak higher, which generally costs low and midrange torque but benefits high-rpm horsepower. Our test of a set of 1 7/8-inch headers compared to a set of 1 5/8-inch headers on a 500hp 406ci small-block netted a gain of 26 hp but lost big torque at the bottom. | 
Intermediate-length headers are also becoming popular, because they are often much easier to fit than long-tube headers. The shorter primary pipe length generally sacrifices some torque on the bottom of the torque curve as evidenced by our test. This was followed by an increase of over 20 peak horsepower at the top end of the power curve. | 
On a small-block Chevy, moving up to 1 7/8-inch primary pipe headers also requires adapter plates, because the larger primary pipes overlap the stock exhaust bolt holes. These adapters use countersunk bolts to mount the flanges with a wider header-bolt flange pattern. |