How to find the right torque spec for any fastener

Every wrenching job comes down to one question: what value do I dial into the torque wrench? Get it wrong by 30% and you can strip a thread, snap a bolt, or — worse — torque something to spec that looks fine and lets go a thousand miles down the road. There is no shortcut to the right number. There is a framework.

This is the framework we use, in the order we use it. It is the same framework that guides how Torq presents specs in the app.

Step 1: Identify the joint, not just the bolt

A “14mm bolt” is not a torque spec. The same physical bolt on two different joints can take wildly different torque values because the joint geometry, the mating material, and the bolt's job are all different. Before you look up a spec, write down five things:

• Vehicle — year, make, model, engine code, trim level. A 2018 WRX and a 2018 WRX STI share a body and lose almost everything else.
• System — engine, drivetrain, brakes, suspension. The system changes which manual section applies.
• Component— “cylinder head bolts” not “the long bolts on top.”
• Fastener identity — thread pitch, length, grade, head type, whether it is a stud-and-nut or single bolt. A spec for an M11×1.5 head bolt does not apply to the M10×1.25 next to it.
• Joint condition — new, reused, oiled, dry. Lubricated threads torque to a different number than dry threads even though the clamp load is the same.

If you cannot answer those five questions, you are not ready to torque anything. Stop and confirm.

Step 2: Go to the OEM service manual first

The factory service manual (FSM) for your specific vehicle is the only source that is authoritative for a given torque value. Other sources — Haynes manuals, Chilton, build guides, forum threads — are derivative. They might be right. They might paraphrase the wrong section. They might be from a different model year.

Where to find an FSM:

• Buy the OEM manual direct from the manufacturer. Toyota, Subaru, Ford, GM, BMW all publish them. They cost between $30 and $200 depending on format.
• Manufacturer subscription portals— Subaru's STIS, Ford Motorcraft, GM Tech2, BMW WebDCS, Mopar TechAuthority. These are typically pay-by-day and are the same documents shops use.
• ALLDATA / Mitchell1 / Identifix — aggregator subscriptions. Useful for working pros across many makes. Less suited to a single-vehicle DIYer.
• Library access— most U.S. public libraries provide free ALLDATA or ChiltonLibrary access through their card. Check your library's digital resources page before paying for anything.

Once you have the FSM in front of you, find the torque-spec table for the system you're working on. Read the conditions surrounding the number: lubricated vs dry, new vs reused, the stage sequence if there is one, and any tightening procedure notes. The number without the conditions is useless.

Step 3: When you can't get to the OEM source, rank what you can find

Sometimes the FSM is unavailable, expensive, or written in a language you don't read. When that happens, you fall back to community sources — but you rank them before you trust them. Our trust ladder, strongest to weakest:

1. OEM technical service bulletin (TSB) — same authority as the FSM and often more current.
2. Manufacturer training materials — dealer technician handouts, factory courses. These reproduce FSM data.
3. OEM-published build/rebuild guides — usually for performance variants (e.g., a Subaru EJ257 short-block rebuild guide from STI).
4. Vetted publications — Engine Builder magazine, Motor Age, SAE papers. Slower but high signal.
5. Specialist shops with documented procedures — IAG Performance for EJ rebuilds, Skunk2 for K-series, Cobb tuning for Subaru. They get sued if they publish wrong data; they keep it tight.
6. Mechanic-driven forums with high engagement— NASIOC, Bimmerforums, Honda-Tech, FordF150Forum. Specifically the threads where multiple working mechanics weigh in and an OP's photo of the FSM page is posted. The thread itself is not the source. The photo of the FSM page is.
7. YouTube — verify the value in the description matches the value in the manual, not just the value the host quotes on camera.
8. Random Google result — never on a torque-critical joint.

Two sources independently agreeing means more than one source repeated by ten people. Provenance over popularity, every time.

Step 4: Sanity-check the number against the bolt

Before you spin the wrench, ask whether the number you have is even physically plausible for the fastener you're torquing. A rough sanity check:

• M6 bolts — typically 7–10 ft-lbs. If someone tells you 25, stop.
• M8 bolts — typically 18–25 ft-lbs.
• M10 bolts — typically 35–45 ft-lbs.
• M12 bolts — typically 60–90 ft-lbs.
• M14 bolts — typically 95–125 ft-lbs.

These are rough ranges for grade-8.8 fasteners with clean dry threads. Cylinder head bolts, stretch bolts, and torque-to-yield fasteners do not fit this rule of thumb. Neither do axle nuts or wheel bearings (which often need 200+ ft-lbs). The rule of thumb catches obvious typos — say, someone wrote 50 ft-lbs where they meant 50 inch-lbs — not subtle errors. Always confirm against the manual.

Step 5: Know when to put the wrench down

Torque-critical work that you cannot source authoritatively is work you should not do. The list of joints where this rule is non-negotiable:

• Cylinder head bolts and main bearing caps
• Connecting-rod cap bolts
• Crankshaft and harmonic balancer bolts
• Flywheel and pressure-plate bolts
• Wheel hub and axle nuts
• Suspension fasteners — ball joints, control arms, sway-bar links
• Brake-caliper bracket bolts and brake-rotor hardware
• Steering rack, tie rod, and pitman/idler arm hardware
• Driveshaft and U-joint bolts

On those joints, you either have the OEM spec with conditions, or you stop. The alternative is a customer driving away on a joint that was never going to hold.

How Torq fits into this framework

We built Torq because looking up specs the right way takes time most working mechanics do not have. The app collapses the framework into a single screen:

• Drill-down by vehiclematches step 1. You can't see a spec until you've told the app what year, make, model, engine, and trim you are working on.
• Trust badges map directly to step 3. Every value carries an OEM, Community, or Unverified badge and links to its source.
• Source citations with page numbers mean you can pull the manual and verify in 30 seconds, not 30 minutes.
• Pro's OEM-only filterhides everything that isn't manufacturer-published. For torque-critical work, you flip the toggle and the screen quiets down.

We are not an OEM. We will never be an OEM. The job of the app is to make the right framework easy to apply. The framework still belongs to you.