A torque spec is only as good as the tool delivering it — and everything between your hand and the nut changes the number. This guide gives you the extension correction formula with a live calculator, the multiplier conversion, honest accuracy expectations for click-type wrenches, the torque-to-angle fallback, and the stripe-marking method inspectors use to prove a joint stayed tight.
Last updated: November 9, 2026 · Maintained by the Bolt Calculator engineering team
This guide is an educational reference built on public engineering practice. It does not replace a calibrated, certificated torque wrench, your equipment’s service manual, or a licensed engineer for safety-critical joints. Brakes, steering, suspension, wheels, and cylinder heads belong to the OEM spec and a tool with a current calibration certificate — if your only option today is a method from the “no torque wrench” section, schedule a proper re-torque with a real wrench as soon as the part is back in your hands.
T_wrench = T_target × L / (L + E)
A straight extension makes the fastener see less than the wrench setting. Divide the target by the length ratio above: with an 18 in wrench and a 6 in extension chasing 100 ft·lb, set the wrench to 75 ft·lb. The calculator just below does the division for you.
No. A click-type torque wrench is a precision setting tool, not a breaker bar. Breaking a rusted fastener loose can demand several times the wrench’s full-scale torque and permanently shifts its calibration. Grab a breaker bar or impact wrench for loosening, and keep the torque wrench for the one job it is certified to do — tightening to a known value.
Both tools update as you type, accept empty fields until you’re ready, and reset with one click. Everything between the wrench head and the fastener that adds straight-line length belongs in the extension box — including crowfeet and crows-foot adapters.
Target torque + wrench length + extension length → corrected wrench setting
100 ft·lb × 18.0 / (18.0 + 6.0) = 75.0 ft·lb at the fastener
Valid for a straight, inline extension under a perpendicular pull — not for flex heads or angled crowfeet. Lengths must share one unit; L runs from the center of your hand grip to the center of the wrench head.
Target output + ratio → required input (or the reverse)
500 ÷ 4 — the wrench clicks at its own smaller scale, the gearbox delivers the big number.
The reaction arm is not optional: it carries the counter-torque and must brace against solid structure. A floating reaction arm lets the multiplier spin and delivers nothing to the fastener.
Four families of torque wrench cover nearly every shop task, and they fail in characteristically different ways. The beam wrench is a spring-steel bar with a scale: there is no mechanism inside to drift, which is why it survives drops and decades of abuse, but the reading depends on your line of sight and where your hand sits on the handle. The click wrench hides a spring-loaded cam that releases with an audible and tactile snap at the set value — the workhorse of automotive and industrial assembly, and the type with the most folklore attached to it. The dial wrench shows torque on an analog gauge, which makes it the natural choice for breakaway testing and stud-torque monitoring where you want to watch a needle climb. Digital wrenches read a strain gauge, add peak-hold, units, and angle modes, and sit at the top of the everyday accuracy ladder.
Selection is mostly about range and resolution, not brand. Work in the middle of a wrench’s scale: a band commonly quoted as 20–100% of capacity means a fastener spec of 30 ft·lb wants a wrench rated around 40–150 ft·lb, not the 250 ft·lb half-inch monster in the bottom drawer. A wrench asked to click near the bottom of its scale spends its tolerance budget on resolution it does not have. Drive size follows from that range — 3/8″ for the sub-80 ft·lb world of engines and brackets, 1/2″ for wheels and suspension, and larger drives or a multiplier for structural work.
| Wrench type | Typical published accuracy class | What moves the number |
|---|---|---|
| Beam (flex-bar) | ±2–4% | Scale reading depends on your line of sight and hand position on the handle; no internal mechanism to drift or fatigue. |
| Click (mechanical) | ±3–4% typical | Quoted band usually applies across the middle of the scale (commonly 20–100% of capacity). Drops and stored-at-full-spring loading shift the trip point. |
| Dial indicator | ±1–2% typical | You watch the needle, so overshot clicks are visible — good for stud torque monitoring and breakaway checks. |
| Digital / electronic | ±1–2% typical | Strain-gauge models add peak-hold and angle modes; battery state and dropping still require periodic calibration. |
Typical published accuracy classes — the bands manufacturers and industry references commonly quote. This is not a certification table: your wrench’s own calibration certificate is the only document that speaks for your tool, and a bench calibrator check is the only way to verify a wrench you suspect has drifted.
Most torque errors are procedure errors, not tool errors. The sequence below is the distilled version of the tightening-methods reference on our homepage, narrowed to the moment the wrench is in your hand.
Look up the value — the bolt torque spec chart lists common grades and sizes — and read the fine print attached to it: dry or oiled threads, new fastener or reused, nut factor assumptions. A spec quoted for lubricated threads over-torques a dry joint by 25–40%.
Dial below the value and approach it rising. Coming down to the setting from above leaves load in the adjustment train and can read a few percent off; many manufacturers state the accuracy band only for an upward approach.
On a click wrench the trip point is set by the spring, not by hand position — but the grip still controls your pull angle. On a beam or dial wrench the hand position IS the calibration: gripping past the handle center adds effective length and silently over-torques.
A steady pull lets the mechanism trip at its designed rate; a jerking motion stores momentum in the joint and lets the fastener overshoot before the click catches up. Keep the pull at 90° to the wrench body — an angled pull shortens the effective lever.
The click is the finish line, not a starting gun. A second 'confirmation' click adds real torque — each re-click continues rotating the fastener while the joint is already loaded. Slow the pull as you approach so the first click happens at low speed.
Back the scale to its lowest setting before the wrench goes back in the drawer. A click spring held compressed for months relaxes and takes its calibration with it — the single cheapest piece of prevention on this page.
Every inch in the drive line changes the numberPhoto: UnsplashBridge: value first, then delivery
Know the number before you correct for the tool: the bolt torque spec chart gives you the target value per grade and size, and this page makes sure the wrench actually delivers it. Look the value up → pull it off correctly.
The wrench does not know what torque the fastener receives. It knows only the force your hand applies and the distance from your grip to the point where resistance enters the drive line. The click mechanism trips at a fixed moment measured about the wrench head — but with an extension in the socket, the fastener sits farther from the trip point, and the same click corresponds to less torque at the threads.
Straight-extension correction
T_wrench = T_target × L / (L + E)
L = effective wrench length (center of grip to center of head), E = extension length added between the head and the socket. The fastener sees the target; the wrench must be set lower by the ratio L/(L + E).
The derivation is one sentence of lever geometry: torque is force times length, the click trips at force × L, and the fastener experiences that same force acting over L + E. Setting the wrench to T × L/(L + E) makes force × (L + E) equal the target — the extension effectively borrows part of the wrench’s trip moment. Note the direction of the error if you skip the correction: the wrench clicks at its full setting while the fastener receives only the L/(L + E) fraction, so an uncorrected 6-inch extension on an 18-inch wrench under-tightens by 25%.
Crowfeet are extensions. A crowfoot adapter effectively moves the fastener away from the head by the distance from the drive square to the center of its open jaw — commonly 1.5–2.5 inches — and the same formula applies, provided the crowfoot is straight in line with the handle. Offset a crowfoot 90° to the wrench body and the geometry inverts: the offset no longer lengthens the lever meaningfully, and the correction is near zero. Measure the actual distance on your adapter rather than guessing.
When not to use the formula: flex-head wrenches, universal joints, and any adapter that bends the drive line change the effective geometry continuously as the joint loads, so a single L + E correction is a fiction. Keep the head locked straight, keep every element inline, and pull perpendicular — the formula’s three preconditions. Also note it cuts one way only: the formula corrects a wrench that is too long relative to the fastener, never substitutes for calibration.
Multipliers: gear ratios and reaction arms
A torque multiplier is a planetary gear box: input torque × ratio = output torque, so 125 ft·lb into a 4:1 unit delivers ~500 ft·lb at the nut — and a normal click wrench becomes a 1000+ ft·lb tool. The catch is physics, not the ratio: every newton-meter the gearbox adds comes back as reaction torque on the gearbox case. The reaction arm must brace against solid, nearby structure positioned so the case cannot rotate. Float the reaction arm and the multiplier body spins; the wrench clicks, the gearbox whirls, and the fastener gets a fraction of the number you calculated. Grease the gears per the manufacturer — dry planetary sets bind and shock-load both wrench and operator.
Computed from T_wrench = 100 × 18/(18 + E). The proportions hold for any wrench — the calculator above runs your actual numbers.
| Setup in the drive line | E (in) | Factor L/(L+E) | Wrench setting (ft·lb) | Wrench setting (N·m) | Reduction |
|---|---|---|---|---|---|
| No extension (direct socket) | 0 | 1.000 | 100.0 | 135.6 | 0% |
| 2 in crowfoot (open-end) | 2 | 0.900 | 90.0 | 122.0 | −10.0% |
| 3 in extension | 3 | 0.857 | 85.7 | 116.2 | −14.3% |
| 6 in extension | 6 | 0.750 | 75.0 | 101.7 | −25.0% |
| 12 in extension | 12 | 0.600 | 60.0 | 81.3 | −40.0% |
N·m column converts at 1 ft·lb = 1.35582 N·m. Note the crowfoot row: a 2-inch adapter alone already costs 10% — the everyday mistake is assuming “small adapter, no correction” while stacking a crowfoot on an extension.
Open a modern head-bolt or rod-bolt spec and you will find something like “tighten to 30 N·m, then +90°, then +90°.” The torque phase only seats the joint; the final preload is set by angle, because rotation of the nut translates directly into bolt stretch, and bolt stretch is what clamp load actually is. Torque alone scatters with friction — up to ±20–30% in dirty or dry conditions — while the angle method sidesteps most of that scatter by using the thread pitch as the meter: an M12×1.75 bolt advances 1.75 mm of stretch per 360°, regardless of how greasy the threads are.
With a mechanical angle gauge — a dial that mounts between the wrench and the socket — the procedure is: torque to the snug value in the spec, zero the gauge, then pull smoothly until the gauge reads the specified angle. Electronic wrenches often have an angle mode built in and do the same job with fewer parts. The known failure mode is pausing mid-pull: static friction grabs, and the resumed pull consumes degrees without stretching the bolt, so commit to one continuous motion per phase.
The free-spinning-nut method removes the guesswork about what “snug” means on a used joint: run the nut down until it stops turning freely against the surface with zero clamp load — a finger-tight condition you can establish repeatably — and treat that as your zero-degree reference. From there, the specified angle produces the specified stretch. This is the standard rescue for joints that were previously tightened and backed off, where the original snug point is buried under crushed coatings and seat wear.
Reading angles without a gauge
Geometry does the measurement for free: a full turn is 360°, and a hex nut has six flats, so one flat = 60°. A 120° finish is two flats — draw a stripe, jump two flats with the wrench position, pull, verify the stripe lands two flats around. A quarter turn is 90°. For the small angles in a first phase (30–60°), a stripe on a nut or bolt head with six marked clock positions gives you a repeatable 60°-per-flat ruler.
Sometimes the wrench is at home and the job is not waiting. These methods are ordered from most to least defensible, and every one of them is a compromise — the amber card at the top of this page doubles for this section.
The best available substitute because it controls stretch rather than friction. Snug the fastener with a short wrench until the washer just bites — no leverage, no body weight — mark a stripe, then rotate the specified angle using flats or a quarter-turn reference. Works for any spec written as torque-plus-angle, and for non-critical joints you can apply a generic snug-plus-90° habit to small hardware. The failure mode is a dishonest snug: if the first bite took effort, your zero is already preloaded and every degree after it over-tightens.
For M6-and-under hardware on brackets, covers, and non-structural parts, finger-tight plus a modest fixed rotation (a quarter to half a flat) lands close enough for the joint to function without stripping. This is how most people successfully reinstall valve covers and plastic intakes. It is not a substitute for a spec — it is a way to avoid the two real failure modes (loose, and stripped) when the spec is small and the material is forgiving.
Torque is force × lever length, so a luggage scale pulled straight down on a bar measures torque directly: 20 lbf at exactly 12 in from the fastener center is 20 ft·lb; 100 N at 0.30 m is 30 N·m. Measure to the center of your grip, keep the pull perpendicular, and the arithmetic is sound — the practice is not: body-height levers amplify small force errors, cheap scales drift 10% or more, and nobody pulls a consistent 90° while squatting at an awkward angle. Treat the result as ±15–25% and reserve it for non-critical fasteners where being slightly loose is recoverable and being slightly tight is harmless. Never on brakes, wheels, heads, or rods.
Short of the methods above: 'wrench-length feel' calibrated on previous jobs, breaker bar + one arm + faith, impact-gun-and-hope. These fail unpredictably in both directions — and an over-torqued fastener is not merely tight, it is permanently stretched, with yield lost before the first re-use.
Verification answers a different question than tightening. You are no longer asking “how much torque should I apply?” but “is this fastener still where I left it?” The basic static re-check: set the calibrated wrench to the specified torque, apply it slowly to the installed fastener, and listen. If it clicks with no nut movement, the joint holds at least the spec. Expect the re-check click to arrive slightly earlier than the installation click — static friction exceeds sliding friction — which is exactly why a re-check is a pass/fail gate, not a re-tightening instruction: if the nut turns before the click, the joint was under-torqued and needs the full procedure, not a half-turn apology.
The torque stripe (or witness mark) turns verification into something an inspector can do from a meter away. After final tightening, draw an unbroken paint-pen line across the nut, across the protruding bolt end, and onto the fixed surface — one continuous stroke spanning all three. Cured paint is the record: any relative rotation between nut, bolt, and surface visibly breaks the alignment of the stripe. Aviation and motorsport practice stripes every safety-wired or safety-critical fastener at installation, because hundreds of joints can then be audited in seconds at every inspection interval.
Be precise about what the stripe proves. It evidences rotation, not torque magnitude — a fastener under-torqued at installation and never disturbed shows a perfect stripe forever. That is why the marking method pairs with the re-check: stripe for the interval audit, calibrated wrench for the periodic proof. On joints with known self-loosening exposure (vibration, thermal cycling), the audit angle matters too — an inspector noting how far a stripe has walked can tell the difference between settlement rotation in the first heat cycle (normal, a few degrees) and progressive loosening (the nut walking its own direction turn after turn).
The click is the measurement — treat it like onePhoto: UnsplashCommon shop practice re-verifies click wrenches every 5,000–10,000 cycles or about annually, whichever comes first — and immediately after any drop, any use as a breaker, or any suspected overload. A verification check against a bench tester takes minutes; a joint failed by an out-of-calibration wrench takes the whole assembly apart.
Breakaway torque on a corroded fastener regularly exceeds full scale and slams the mechanism past its stop. One stuck bolt can retire a wrench. Breaker bar for loosening, always.
The lowest-setting storage rule exists because a compressed spring relaxes over months and the trip point moves with it. Dial to minimum before the drawer, every time.
Crowfoot + extension + universal = the formula no longer describes your joint. Keep it straight, keep it inline, correct with L/(L+E), or lose the accuracy you paid for.
Each extra click keeps rotating a loaded joint, and a jerky pull lets momentum overshoot before the mechanism trips. Slow pull, first click, done.
The same spec on dry vs oiled threads delivers wildly different clamp load. Read the spec's lubrication condition — or dial the value for the condition you actually have.
A drop onto a concrete floor is a calibration event even when the wrench looks fine. Re-verify before the next critical joint, not after the failure investigation.
A 150 ft·lb wrench clicking at 15 ft·lb spends its ±4% band on resolution error many times larger. Choose the wrench so the job sits in the middle of its range.
The seven questions this guide gets asked most — from loosening to stripes to how much to trust a click.
This guide makes the wrench honest, but the target still has to come from somewhere defensible. Pull the spec for your bolt grade and size from the bolt torque spec chart, correct it here for whatever is in your drive line, then verify with the re-check and stripe method above. For the full method family — hand methods, angle control, and when each is appropriate — the tightening-methods section on the homepage is the overview, and the guides hub collects every reference page on the site.
The wrench is the last step — these companions cover the number and the joint behind it.
Maximum allowable torque for AN3–AN20 aircraft bolts per FAA AC 43.13-1B, with fine/coarse and dry/lubricated columns, NAS/MS families and self-locking nut notes.
Electrical connection torque chart per UL 486A-486B and NETA ATS practice: lugs, terminals and busbars from #8 to 1/2 in and M5 to M12, with cold-flow notes.
U-bolt torque values by diameter and thread for leaf springs and trailer axles, with staged tightening, re-torque practice and axle-rated recommendations.
Head bolt torque specs, tightening sequence patterns, TTY identification and reuse rules, and re-torque guidance — organized by application, not bolt size.
Metric and SAE thread pitch chart: pitch and TPI, major, pitch and minor diameters, tensile stress areas As, and tap drill sizes for M2–M48 and #0–1-1/2 in.
Hex head and hex nut dimensions for M5–M30 and 1/4–1 inch sizes: across flats, head height, nut thickness, heavy hex and SAE–metric conversion.
ISO 273 and ASME B18.2.8 clearance hole charts for M2–M30 and #4–1 inch screws, with fit selection steps and floating/fixed fastener checks.
Metric and SAE bolt torque spec charts: Class 8.8/10.9/12.9 and Grade 5/8 values dry and lubricated, plus B7 stud, B8 stainless and SHCS rows.
Convert torque between N·m, ft·lb, in·lb and kgf·m instantly, with a 10–200 N·m quick chart for bolt specs. Free, in-browser, no sign-up.
Bolt size calculator from load: enter force, safety factor and grade to get the minimum standard diameter, tensile stress area and utilization check.
Bolt thread engagement calculator: stripping force vs bolt fracture, minimum engagement length for steel, aluminum, brass or cast iron tapped holes.
Flange bolt torque calculator with hydrostatic end force and gasket seating: per-bolt load, T = K·F·d torque, scatter band, and % yield target — free.
Eccentric bolt group calculator: elastic center, polar inertia, per-bolt shear and moment forces, worst-bolt utilization and pass/fail — free and in-browser.
Estimate concrete anchor capacity: breakout tension from embedment, F1554 rod strength, projection above the slab and J-bolt weights — ACI 318 style.
Work out bolt length from grip, nut height and protrusion: through-bolt, A325 structural, TC bolt and flange stud scenarios, rounded to standard sizes.
Get X/Y hole coordinates from a bolt circle: BCD from chord or across-holes measurements, hole angles, uneven spacing — flanges, wheels, chainrings.
Check bolt shear stress with interactive sliders: bolt count, diameter, single or double shear, grade-based allowable stress and pass/fail safety check.
Weigh a complete fastener set: hex bolt plus nut and washer options, per-piece and total weight by quantity. Free, instant, no sign-up.