Chapter XIII
Splines & Driveline Lubrication
This is the chapter that saves gearboxes. Everything else in this book is a number you look up once; this is a job you have to keep doing, on a clock nobody at BMW ever wrote down. The factory service schedule does not mention the transmission input splines at all. They are, nevertheless, the single most consequential scheduled item on an airhead — and the failure they produce is expensive, unheralded, and entirely preventable.
Be careful with the popular framing. There is no canonical "five spline points" list for airheads; that list belongs to K-bikes and later machines. A twin-shock airhead has three greased spline interfaces plus one that lives in oil. A Monolever or Paralever has fewer still — no rear wheel splines at all. The table below separates the points you actually service from the ones you only ever touch during assembly, because conflating them is how people end up disassembling something that needed nothing.
The deeper disagreement is worth stating up front, because it changes what you conclude when you find wear. One school holds that dry, rusting splines are the cause: grease them on schedule and they last. The other observes failures at 20,000 and 30,000 miles with the grease still intact, and points instead at misalignment — a distorted clutch disc, a rocking hub, a bent housing. Both are documented, by serious people. If you find heavy wear on a well-greased spline, do not simply re-grease it and reset the clock.
Every figure on this page resolves to one model and year in the book. Pick your bike and the rows that apply to it come forward, with the rest marked.
Open this chapter in the bookThe service points
| Location | Interval | Access | | Applies | Consequence of neglect |
|---|
| Transmission input (clutch) splines | 15,000 mi | gearbox slid back — engine stays in | do not exceed 18,000 | All to ~1993/94 | The one that matters. Roughly double the interval on later nickel-plated shafts, though the plating date is disputed across sources (1985 / 1989 / 1993–94). Neglect ends in stripped splines, a stranded bike and a full gearbox overhaul and re-shim. |
| Rear wheel cup ↔ rear drive output splines | every tyre change | rear wheel off | inspect the cup rivets first | Twin-shock only | Monolever and Paralever do not have these. Brush moly grease from the outboard end inward, so grit is not driven into the large drive seal. The drive-side splines wear much faster than the wheel-cup side. |
| Driveshaft rear splines at the rear-drive pinion | with the housing oil, 10,000 mi | normally never serviced | | oil-housing models | These live in the driveshaft-housing gear oil — if the housing is correctly filled, the oil is above the entire splined coupling. Keep the oil in it and leave them alone. Paralever uses no oil in the shaft housing and is checked at the 15–20,000 service instead. |
| Clutch actuating lever, pin and bush | every 2nd spline service | rear of the gearbox, between the ears | | All | Clean and re-grease. See the pin safety upgrade below — this is a genuine stranding risk, not housekeeping. |
| Throwout bearing | at every spline service | behind the rubber bellows cup | non-moly grease only | All | It gets no gearbox oil for some time after reassembly, so it needs its own light, smooth, non-fibrous, non-moly grease. Do not remove the inner pushrod ahead of it. |
| Swingarm pivot bearings | grease ~2×/year · service 15–20,000 | grease gun into the pivot adjuster | full clean every few years | All | Fill the cavity between swingarm and adjuster boss completely and smooth it round with a finger. Documented failures are rusted and cracked bearings, and brinelled outer races. |
| Steering head bearings | 30,000 mi | drop the forks ~1 in | sources also say 50,000 / 4 yr | All | Doable with the fairing in place. Cleaning and regreasing frequently cures notchiness that appears to demand new bearings — a shop will not attempt it because labour cost forces straight replacement. |
| Wheel bearings | every tyre change | tapered 30203-A · 17×40×12 | pre-1985 only | pre-1985 | 1985+ are sealed 6005 balls — replace, do not service. R65LS went to balls earlier. Never moly here — it clumps and flakes out. |
| Speedometer drive boot | inspect yearly | right rear of the gearbox | | All | Airhead speedo drive is at the transmission, not the front wheel. Water in here turns the gearbox oil to creamed coffee. Seal with black RTV; no zip ties — they squeeze the rubber out of round. |
| Rear brake cam O-rings | every 2nd tyre change | silicone grease | | drum rear, to 1984 | From 1985 the rod runs in a tube with no O-rings — grease the shaft anyway. |
Assembly-only points — not maintenance items
| Location | Treatment | | | Applies | Notes |
|---|
| Transmission output flange ↔ output shaft taper | clean and DRY | flange nut 148–173 ft-lb | no grease, no oil | All | Not a wear point. The documented failure here is the four U-joint bolts backing out and tearing the rear off the gearbox — check them yearly at 26 ±2 ft-lb without loosening first. |
| Rear drive input (pinion) nose splines | sealant, not grease | plastic deform-seal post-/5 | | All | Oil migrates down these splines into the drive nose. On /5, non-hardening Form-A-Gasket on the forward half-inch of splines. |
| Flat forward end of the input shaft | no grease | one small drop on the pushrod tip | | All | Grease on the flat end just gets thrown into the clutch later. |
What to use where
| Point | Lubricant | Named products | | Applies | Notes |
|---|
| Transmission input splines | moly grease/paste | Guard Dog GD525 (30%) · Honda M60 · Honda M77 / Molykote M77 · Loctite 234227 · Molykote U-N (>60%) | a moderately thin worked-in layer only | All | Moly percentage is disputed — see the conflicts below. Second tier: Chevron Ultra Duty EP with moly added, Würth SIG3000 with moly added (SIG3000 alone contains none), Texaco Starplex-2 with Moly, Mercury Marine 92-816391R4. |
| Clutch disc female hub splines | the rule is: do not | — | see conflict below | All | The published instruction is emphatic — never grease the disc splines. The same author later admitted to using a very tiny amount, but only with the gearbox fully out. On the slide-back procedure you cannot clean the disc properly anyway, so don't. |
| Throwout bearing | light non-moly grease | Chevron Ultra Duty EP | never moly | All | Low-viscosity, smooth, non-fibrous. |
| Clutch actuating arm bearing | oil, not grease | moly grease only on the outer rubbing faces | | All | Needle-bearing type: soak in warm kerosene, dry, warm, oil. A droplet of moly at the rod contact point. |
| Rear wheel & drive splines | any decent moly grease | — | | Twin-shock | Brush outboard-to-inboard. |
| Wheel bearings | non-fibrous EP grease | Chevron Ultra Duty EP NLGI 1 or 2 · StaLube SL3161 · Quaker State Multipurpose | NEVER moly | pre-1985 | Moly clumps and flakes out in rolling-element bearings. |
| Swingarm & steering head bearings | soft non-fibrous grease | ± a little moly — disputed | NLGI 0 not acceptable | All | One page says mix a small amount of moly in; the same author's steering-head tip says use no moly at all because it clumps and causes notchiness. Both positions below. |
| Cable barrel ends and pivots | moly-containing oil | or plain 20/30 wt non-detergent | not 3-in-One, not WD40 | All | Lubricate often, especially after washing. The plastic clutch lever pivot bushing is not to be lubricated. |
| Exhaust port finned nut threads | nickel anti-seize | Never-Seez Pure Nickel Special · Permatex · Copper-EZE | never oil | All | Oil carbons up and seizes them. Brass-brush the threads first, and stop immediately if a nut starts to seize on removal. |
| Gearbox · driveshaft housing · rear drive | 80W-90 GL-5 | petroleum, not synthetic | | All | See Capacities & Fluids, and the GL-5 vs GL-4 dispute in Sources & Conflicts. |
Do not mix greases. Incompatible bases can cause problems, and the practice of thinning a moly paste with lithium grease so it "stays put" — once a common shop tip — is now explicitly warned against by the same source that used to make specific mixing recommendations.
The 1981 change — three separate things, usually conflated
| What changed | Detail | Effect | | Applies | Notes |
|---|
| Input shaft shortened | ≈24 mm → ≈19 mm | less engaged spline length | | 1981+ | The strongest factual basis for "1981+ is weaker" — less contact area for the same torque. Note that no source states that causal conclusion outright; it is inference from the dimensions. |
| The 1981-only clutch disc | 21-21-1-242-370 | superseded by 21-21-1-451-512 | soft splines · thin steel plate | 1981 only | This is the actual documented culprit. Soft disc splines and a too-thin steel plate that flexed under load, wearing away the mating input shaft splines. Carrier went 21-21-1-242-372 → -451-511 → 21-21-1-338-722; early 6-ear carriers twisted, later full-circle ones much less. Factory washers fitted between carrier and cover on some early 1981 bikes are not to be reused. |
| Clutch and actuation overhauled | flywheel → lighter clutch carrier | rear input seal 23-12-1-243-522 | | 1981+ | Pre-1981: felt ring 23-21-1-230-440, pushrod in from the front. 1981+: lipped seal at the rear cover, rod installed from the rear, rotating, or you destroy the seal. Replacing that seal means splitting the gearbox. Lighter lever effort, more engine vibration. |
| Throwout bearing | needle → ball | from September 1980 | piston OD must not exceed 28.7 mm | Sept 1980+ | The 5-speed's 1974–1980 flat radial needle type is the poor design. Oversize pistons stick when hot, causing slip or stiffness — measure, and sand if necessary. Do not try to convert between styles. |
| The 1982–85 gear problem — a different failure | helical angle 15° → 17.5° | ≈ trans serial 58225, Mar/Apr 1982 | shock coupler ears break off | 4/82 – 4/85 | Frequently mis-dated to 1981. This is not spline wear at all — the drive ears break off the input shaft gear. Fixed March/April 1985 from serial 115167 with input shaft assembly 23-21-2-302-331 (no kickstarter) / -332 (with). 15° and 17.5° gearsets do not interchange; 17.5° is marked with a cross or X. |
Interchange
| Question | Answer | | | Applies | Notes |
|---|
| Do pre-1981 and 1981+ input shafts interchange? | No, not directly | shaft length differs | | All | 24 mm vs ~19 mm. |
| Early gearbox into a 1981+ bike? | Yes, with work | different input shaft + re-shim | | 1981+ | Cheaper documented workaround: shorten the earlier input shaft splines to suit the 1981+ clutch — doable with a cut-off disc with the gearbox still assembled. Radius the forward edge properly. |
| Later gearbox onto an earlier bike? | Yes | with matching carrier & flywheel changes | | All | Watch input shaft length and neutral switch type; Monolever and Paralever versions differ. |
| All 4- and 5-speeds 1970–1980 | Interchangeable | — | clutch actuation is not | 1970–80 | The entire clutch actuation mechanism differs between 4-speed and 5-speed. |
| Overhauling a 1981 bike on original parts | replace everything later updated | — which is nearly all of it | | 1981 | Doing a 1981 clutch with 1981 parts is rebuilding the problem. |
Inspection — what wear actually looks like
| Stage | Input splines | Rear drive / wheel cup | | Applies | Notes |
|---|
| Healthy | grease intact, no discolouration | square outer edges | | All | The trap: input splines are small, thin and slightly pointy-looking even when brand new. Do not condemn on shape alone. |
| Due for service | a rusting look | — | | All | Clean and re-lube before rust appears — rust means wear is already happening. |
| Contaminated | oily spline, oil at the input end | — | | All | Input seal has failed. Pull the gearbox and replace it, or it will ruin the clutch. |
| Badly worn | mostly worn away, distinctly pointy | under ~1/3 of the square outer edge left | rebuild now | All | The rear drive is the only place with a stated numeric rule of thumb: when less than about a third of the original square outer edge remains — i.e. two-thirds gone — it is time to rebuild. |
| Imminent failure | very uncomfortable noise, then no drive | totally pointy, no square edge left | | All | There is no second warning after the noise. |
| Misalignment signature | wear at the very inside of the splines, not along the length | hub rocks on the shaft | discoloured pressure-plate faces | All | If you see this, re-greasing and resetting the clock is the wrong answer. Look for a distorted disc — commonly caused by letting the weight of the gearbox rest on the clutch disc hub during assembly. |
| Also inspect | — | wheel cup rivets and rivet holes | | Twin-shock | Loose rivets and wallowed holes, usually from hard takeoffs. Rebuilding drive splines without a new cup is false economy if the cup shows more than modest wear. |
Symptoms before failure
| Symptom | Points to | | | Applies | Notes |
|---|
| Downshifting gets progressively worse | dry or rusting input splines | the classic early tell | | All | Long before anything breaks. |
| Hard to get 2nd down to 1st, or to engage 1st | input splines need lubrication | — | | All | Specific and reliable. |
| Clutch feels grabby, shifting stiff | dry input splines with rust | — | | All | |
| Shifting turns notably silky after a service | confirms it was the splines | — | | All | A useful retrospective diagnosis. |
| Free play felt at the rear wheel rim | not splines — the large cardan bearing | shop check ≈0.020 in at the rim | not a factory limit | Monolever · Paralever | Once past it, wear accelerates. The figure is shop practice — BMW's late manuals call for correction when play is perceptible at all. |
| Some clutch spline chatter at idle | normal | — | | All | Along with assorted transmission rattles. Not a symptom. |
There is no measurable wear limit for the input splines. No source gives a tooth thickness, a wear depth, a percentage or a go/no-go gauge. The only published test is visual, and the authors concede it takes experience. The rear drive is the one place with a number, given above. That is a real gap in the literature, not an omission here.
The clutch actuating pin — a stranding you can prevent
| Item | Part | | | Applies | Notes |
|---|
| Flanged "top-hat" pin upgrade | 23-13-1-241-484 | clip 51-23-1-864-963 | | 1973 → Sept 1980 | The mid-era pin is held by a C-clip on the inside top of the lower ear. If the clip drops out and the pin walks, the very next pull of the clutch lever breaks off one of the two transmission ears and strands you. The flanged pin cannot fly out. Note: an incorrect number, 23-21-1-241-484, circulated for years — the correct one is 23-13. |
| The clip, honestly | 51-23-1-864-963 | superseded by 07-12-9-934-915 | catalogue fitment does not list airheads | — | Both leading authorities cite this number for this airhead job, and the part is real — but the catalogue applications shown are K-bikes and cars, and it is marked end-of-production. Order it, but expect to have that conversation with a parts desk. |
| Documented alternative | shoulder bolt + Nyloc nut | moly grease on assembly | do not squeeze the boss ears together | All | They break. Tighten only until the bolt still has a little free play. On 1981+ bolt-type installations, fit the bolt nut-down. |
| Pin design by era | /5 cotter pin · mid C-clip · late 10 mm bolt | late arm has needle bearings, no zerk | | All | The /5 cotter-pin design gave no trouble — bend at least one leg over and replace it after a couple of uses. The early arm had a grease nipple. |
Torques you meet on the way in
Before you touch the flywheel or clutch carrier, block the crankshaft. If it moves, thrust washers can drop out of position and the damage is serious.
What it costs to ignore
| Failure | What breaks | The repair | Applies | Notes |
|---|
| Input spline failure | input shaft splines stripped | clutch disc hub usually destroyed too | All | Engine and gearbox split, clutch out, and typically the whole transmission overhauled and re-shimmed. No verified airhead-specific cost figure exists — the four-figure numbers circulating on the forums are oilhead R1150 jobs and do not transfer. |
| Rear drive spline failure | crown wheel output splines and/or cup teeth lost | specialist cut, weld and re-machine | Twin-shock | Rebuilding is typically many hundreds cheaper than a new matched gear set plus the labour to re-shim it. Reported quotes run around $150 for a spline rebuild against roughly $500 for new pinion and crown. The assembly must stay undistorted and concentric. |
| Input seal ignored | gearbox oil down the splines | clutch soaked and ruined | All | The gearbox must come out — but need not be disassembled — to replace the seal. |
| U-joint bolts loosen | flange and U-joint part company | the rear of the gearbox can be torn off | All | First-hand dealership account. This is why the yearly bolt check is on the schedule. |
| Loose rear drive input nut | rapid failure of the drive | — | All | Check it. |
| Water via the speedo boot | emulsified gearbox oil | — | All | A very high repair bill for a five-minute inspection. |
Conflicts on this subject, recorded. Root cause: lubrication failure versus misalignment and disc flex — the most substantive technical disagreement in the whole airhead canon, and the one that should change how you interpret wear you find. 1981 root cause: two accounts appear on the same page, one from jig-testing returned housings concluding the clutch plate steel was too thin, the other claiming some 1981 castings were dimensionally wrong and unfixable. Moly percentage: ~30% is called adequate, while others insist on 60–70% pastes — and the 30% advocate's own favourite product is over 60%. Anti-seize as spline lube: nickel anti-seize (Never-Seez Pure Nickel Special) is described as possibly excellent on the strength of a single report; forum consensus rejects anti-seize outright. Moly in steering head and swingarm bearings: yes-a-little versus absolutely-not, from the same author on two different pages. Clutch disc splines: never, versus a very tiny amount with the gearbox out. Nickel plating start date: 1985, 1989, 1993 and 1993–94 all appear in print; treat it as unresolved and use the shorter interval. Rear drive input nose nut: 110–115, 120, and "150 Nm or 110–120 ft-lb" — use 110–120.
Chapters