A recip-saw blade may not have to work that hard in new construction, but on a remodeling site it undergoes a real torture test, cutting abrasive roofing, nail-embedded wood, and anything else that stands in the way.

Many tool companies produce one or more blades designed for this kind of work. To see how the various choices stack up, I tested blades suitable for cutting nail-embedded wood and doing general demolition.

Blades of this type usually have six teeth per inch (tpi), though some have fewer and others have variable spacing. All but two of the blades

I tested were bimetal – hardened steel tips on a softer steel body. The other blades were carbide-tipped specialty models.

For ease of testing I stuck with 6-inch models – though many of the same blades can also be found in 9- and 12-inch lengths. Standard recip saw blades are .035 inch thick; I tested thicker models, without distinguishing between .050-inch blades and the thicker .062-inch models typically referred to as demolition blades. Both will cut the same things – one is just a little stiffer.

Since my focus was on tooth wear and cutting speed, I counted and timed cuts while running the blades
to destruction.

Test Planks

My first challenge was to come up with something to cut that would wear blades out quickly enough to separate the top performers from the also-rans. Yet I also wanted the test to be as realistic as possible, so I chose not to use stainless steel, hardened fasteners, or cement board, because those materials are not often cut during demo work. Nail-embedded wood. After much trial and error, I came up with a test plank of two 2x6s on edge glued up with a layer of OSB in between and capped with one layer each of drywall and OSB.


Each 8-foot plank contained a dozen rows of nails – 96 feet in all – laid end-to-end in kerfs cut into the 2x6s: ten rows of 16d commons, one of 16d sinkers, and one of little 8d sinkers that proved to be the assembly's secret weapon. By locating the nails along both sides of the central piece of OSB, I was able to concentrate wear on a limited number of teeth and hasten the destruction of the blades.

Building the planks was time-consuming but worthwhile, because it created a difficult but realistic test that would challenge every blade equally.

Simulated roof. Since cutting openings for skylights and vents is a common remodeling task, I designed a test plank to mimic that operation on a three-layer roof. The blank consisted of six asphalt shingles sandwiched between two 6-inch rips of 7/16-inch OSB (the second layer of OSB held things together). To avoid having to cut wide of any connecting fasteners, I glued the planks up with thin beads of polyurethane construction adhesive.


The test plank consisted of a piece of OSB sandwiched between 2x6s and capped with drywall and more OSB. The central piece of OSB was flanked by 12 rows of nails – so the recip blade had to cut 12 nails each time it went through the plank.


For timed cuts, the plank was clamped into a test rig and cut with a saw attached to a pivot arm. The nose of the tool was weighted to provide downward pressure – approximately 23 pounds at the blade.

Test Rig

It's easy to count cuts but difficult to accurately time them while cutting by hand, so I had to build a test rig.

The rig consisted of a woodworking vise for holding the test plank and a sturdy pivot arm for holding the saw. Both were bolted to the thick LVL top of a worktable. For the test saw, I chose the most powerful Milwaukee Sawzall, a 15-amp model with a 1-1/4-inch stroke. To ensure adequate feed pressure, I strapped the weights from a 25-pound dumbbell to its nose. The feed pressure measured 23 pounds at the center of the blade.

During testing, I cut at high speed in nonorbital mode; the rig was so solid it transferred nearly all of the saw's energy to the blade. There was none of the shaking and vibrating you get when holding a saw.


The author used a stopwatch to time cuts through the test planks. A blade was considered dead when it took more than 37.5 seconds per cut; by that point, most blades could barely get through and were hot enough to char the wood.

Cutting Nail-Embedded Wood

At first I planned to make all the cuts in the test rig, but repositioning the test plank for every cut proved to be very time-consuming. Also, the blade sometimes wandered out of the end of the plank when I made thin cuts. To avoid these problems, I alternated between timed cuts in the rig and untimed cuts with the same blade in a hand-held saw.

I timed cut numbers 1, 11, and 21, and – as the blades began to wear – every fifth cut after that (26, 31, and so on). I made the cuts in between by hand. After every 20 cuts I rested the blade and went on to the next model. I continued in this manner until each blade failed.

When is a blade dead? My original benchmark for failure was 45 seconds; the blade would be considered dead if it could not make it through the plank in that amount of time. I came up with that number after speaking to manufacturers about the way they test blades. But in my testing, most blades failed rapidly at times between 35 and 40 seconds, so I adjusted the cutoff down to 37.5 seconds – the halfway point between those numbers

It was pretty obvious when blades failed – they smoked, sparked, and ceased to make progress through the nails. In all cases, I stopped timing if a blade couldn't complete a cut in one minute and 30 seconds.


Tooth size alone is not a good predictor of longevity. The widely spaced teeth of the Ridgid blade (at top) cut quickly at first but were prone to snagging on nails. The more tightly spaced teeth of the Starrett didn't lose teeth to shock, but friction and heat eventually wore them down.


Blades cut faster and last longer when used in orbital mode. These DeWalt blades were tested to failure; the upper one made 53 cuts in linear mode and the lower one 100 cuts in orbital.