Wire harness failures: what 422 UK vehicle recalls show

Of the 4,566 vehicle recalls launched in the UK from 2016 to 2025, we count 422 where wiring, a connector or an electrical joint is the stated problem.¹ The largest group, 127 recalls, is routing: a harness too close to something, or rubbing on it. Next come bolted joints left loose or under-torqued, at 74. Terminal and joint faults come to 19, and the word crimp appears in only 2 of the 422. Recall text rarely says how a joint was made, so that is a limit of the record and not proof that crimps are sound.

Count vehicles instead of recalls and the order changes. Fretting, wear at a contact from small movements, is only 6 recalls but 21% of the vehicles listed. Bolted joints are 1%, with a median of 22 vehicles per recall. A vehicle count is the number in scope, not the number that failed.

Where the text lets us place the fault, we read 144 of 193 as made while building the vehicle, 33 as faulty parts and 16 as design. Few say so outright: 36 of the 144 use a word like assembly, production or installation about the fault itself. The rest say that a joint may not be tightened or a harness may not be routed correctly, and we counted that as assembly. Of the 33 faulty parts, about 15 are the harness or its own connector.

Three questions come out of this on our reading. Can this harness be fitted wrongly? Does anything record that a high-current joint was tightened? What stops the harness moving its connector?

What failed, and how often

Fire is named as a possible outcome in 56 of the 422, and 8 more say thermal event or thermal incident. Routing accounts for 20 of the 56 and bolted joints for 10. Water and corrosion reach 7 out of only 23 recalls, a high share for a small group.

By vehicles listed, the largest groups are fretting (21%), earth path (20%) and overheating (14%). Routing falls to 8%.

Routing: what the harness ran into

We read each of the 127 routing recalls for what the wiring met.

The most instructive is Honda's R/2018/225, covering 5,097 vehicles. In the recall's words: "During vehicle assembly the cabin harness was pulled to make the process easier to manage." Pulling broke or dislodged a clip, and in around 10% of affected units the harness touched the transmission mount, where vibration could damage it.

Our reading is that a harness awkward to fit will get pulled, and that no inspection of the finished harness can see that coming.

Seats earn their own line because the fault may exist at only one end of their travel. Aston Martin's R/2017/343 (6,542 vehicles) arises when the right front seat is run fully back, where the seat runner can compress the battery supply cable. On our reading, a check with the seat parked mid-way may well pass.

The wiring can also be what does the damage. Ford's R/2024/503 describes a clash between a harness and the brake lines that could lead to wear, corrosion and a brake tube leak. Vauxhall's R/2024/469 has a fuel pipe chafing in contact with the engine harness. Mercedes-Benz R/2021/412 (12,696 vans) is simply about too little distance between the brake line and the harness.

These recalls report the fault. The checks below are our suggestions for catching it earlier.

• Walk the route with every moving neighbour at both ends of its travel: seats forward, back and folded, pedals to the floor, steering lock to lock, doors and tailgates, the engine rocking on its mounts.
• Put minimum clearances on the drawing for brake and fuel lines, exhausts and moving parts. "Route as shown" leaves the fitter to guess what matters.
• Ask the fitters where they have to pull. Then change the slack, the fitting order or the clip.
• Count clips after fitting, including the ones that take a mirror to see.
• Give hand-fitted edge protection its own check. Volvo Bus R/2024/256 traces chafing power cables to edge protection that may not have been fitted correctly.

Bolted joints: the group that is growing

Bolted-joint recalls went from 19 in 2016 to 2020 to 55 in 2021 to 2025, or from 0.85% to 2.4% of all UK vehicle recalls. Those 55 come from eleven manufacturer groups, with Mercedes-Benz and Daimler at 20, so loose joints are more than one company's habit. The 48 V recalls are narrower: all 27 come from Mercedes-Benz and Daimler (12), BMW and MINI (7), Ford (6) and Maserati (2), and all but one date from 2021 or later.

Most are small, with a median of 17 vehicles, and 23 of the 55 list 10 or fewer. That could mean more loose joints, or better torque records that find a few affected vehicles. The text does not say which. Of all 74, 62 describe a joint not tightened or fitted to specification, which we read as an assembly fault, though only 4 of those mention production or manufacture.

Small extra resistance matters far more at high current, because heat in a joint rises with the square of the current.

We calculated these from P = I²R and V = IR. The currents are round figures to show the scale. They are not from any recall, and we would expect 300 A to be a brief peak in most vehicles.

On our reading, a loose joint escapes for a simpler reason: the harness test happens before the joint is made, and a continuity limit sits far above these values. A 4-wire meter at 1 A can resolve 5 mV, but a joint with too little clamp force can still read low on the day and rise later.

The records also show what ends up where metal should meet metal. Ford's R/2023/204 is tape trapped beneath the nut on a 48 V battery earth. MG's R/2023/050 (3,291 vehicles) is insulation trapped between an earth wire and the body. Rolls-Royce's R/2024/022 is adhesive residue on a cable eyelet. Fiat's R/2024/434 goes further: the nut on a high-voltage battery terminal may be missing. The remedy replaces the whole battery, so this one sits at the edge of our scope. Mercedes-Benz R/2016/032 blames "deviation in the tightening angle monitoring", which we read as the monitor itself being part of the problem.

If you make high-current joints in anything, a vehicle or a panel, we would start here:

• Record torque and angle for each joint against a serial number, and audit the monitoring system itself.
• Keep the eyelet and its seat visible and clean when the nut goes on. Tape ends, adhesive and stray insulation are all on the record above.
• Use a 4-wire reading as a second check after recorded torque, not as proof. A continuity test cannot see this fault.
• Treat panel busbars and terminals the same way, and make them part of routine verification.

Swapped pairs can pass a harness test

Wrong wiring accounts for 24 recalls, and at least ten are swaps of two circuits. Jaguar's R/2017/197 (8,987 vehicles) has the right C-pillar crash sensor wired to the left input and the reverse. Mercedes-Benz R/2017/199 swaps left and right airbag firing "due to an incorrect connector assignment on the wiring harness". Iveco's R/2025/195 may invert the ABS connections to the front wheels. Dennis Eagle's R/2021/300 (1,318 vehicles) may have ABS plugs in the wrong modulators. Mercedes-Benz Vans R/2022/069 may swap the alternator harness with the electric steering harness.

A harness tester compares each harness with its netlist. If a harness is built wrongly against a correct netlist, the tester should catch it. If the netlist carries the swap, every unit passes. If a correct harness is fitted the wrong way round, it has already passed. No recall says which of these happened, so this is our reading.

Nine of them show where the swap sits. Five are pin errors inside one connector: Jaguar R/2017/197, Mercedes-Benz R/2017/199, Rolls-Royce R/2019/094, Bentley R/2021/113 and Citroën R/2025/142. Keying and colour cannot prevent those. Check the netlist against the module's pin-out at design review, and check wire colour or position at pin insertion. The other four are whole cables or plugs in the wrong place, where keying, different stud sizes or branch length can rule the mistake out.

Test the finished product by function where you can. Airbags and pretensioners cannot be fired on the line, so use the control unit's diagnostics or a stand-in load, and ask how that test tells left from right.

Six recalls, and the two largest in the set

Fretting is named or clearly described in only 6 recalls, but those include the two largest of all 422 by vehicles listed: BMW R/2018/132 (311,848) and BMW R/2018/148 (279,104). The third largest, Vauxhall R/2019/082 (234,931), describes connector movement and heat-damaged terminals without naming wear, so we count it as overheating and treat fretting there as likely, not stated. BMW's earlier R/2017/114 (36,410 vehicles) has the same concern title as R/2018/132 and says wear and heat but not movement, so we count it with terminals.

BMW's R/2018/148 sets out the chain: irregularities in the crimp, relative movement of a tin-coated connector on a tin-coated pin, frictional corrosion, high current, temperature rise, damaged wire strands and a fire risk. Toyota's R/2016/184 has tin plating abraded by a sliding terminal until oxidised debris builds up. Then a driver's knee presses a lower cover against the harness, the terminal slides onto the debris, contact fails and steering assist stops.

On our reading, a crimp height reading or a pull test on the day of manufacture would not show the wear itself. The crimp can still matter: BMW lists irregularities in the crimp as the first link in its chain.

Five recalls concern one job: BMW R/2018/148 and R/2019/192 on the blower regulator connector, BMW R/2021/064 and Alpina R/2021/408 on the blower regulator wiring, and Vauxhall R/2019/082 on the fan speed resistor connector. Together they list about 729,000 vehicles, 25% of all those listed in the 422, though not all of them name wear. If you have a connector carrying blower current, start there.

The recalls do not agree on one remedy. Vauxhall added "additional mounting points" and a redesigned connector plug. Chevrolet R/2016/175 replaced a battery tray and added contact sealant, because the movement came from the ECU bracket and tray as well as the harness. On our reading, fix the harness and the connector to the same moving body, stop the part the connector sits on from moving, and ask the connector maker what the contact system is rated for under vibration, including contact force, plating and lubricant. Do not limit this to power circuits: neither Chevrolet nor Toyota says the terminal carried high current.

The smaller groups

How the count was made, and what it cannot tell you

We started from the DVSA recall data file, which lists 4,566 recall numbers launched from 2016 to 2025. Three word searches for wiring terms gave 863. The first 663 and 27 of the rest were labelled one by one. The other 173 matched only on words such as connection, turned out to be threaded joints, fuel and brake connections and similar, and were labelled out of scope as a block. A recall counted if the defect lay in wiring or in how it was made or fitted. Failures inside a sealed unit such as a battery module or a control unit did not count, even when the text used the word terminal or cable. Where a harness and a fuel or brake line rub, we counted the recall whichever part it says is damaged.

The reading and labelling were done with AI assistance, and no engineer has checked the labels. After the first pass we moved 75 recalls one by one into more specific groups, adding groups the first pass did not have, and removed 10 as out of scope. A second AI pass, using the revised instructions, then labelled a random 100 of the first 663 without seeing the labels. It agreed on 95 of the 100 for scope and on 64 of the 67 that both counted in scope for type of failure. Against the first pass, before the moves, that figure was 52 of 67. The sample holds only a few recalls from the smaller groups and none of the 200 found by the later searches. After that check and a separate AI review of this article, we removed 9 more recalls and added 23, 18 of them from the later searches.

Keep these limits in mind.

• Nothing here says how often anything fails. About half of these recalls list fewer than 300 vehicles.
• DVSA's guide says each row of the file repeats the recall total, which fits the file up to 2022. From 2023 many recalls show different numbers on different rows. We used the highest row, which may understate 34 of the 422. Adding those rows instead leaves the half under 300 and the two largest unchanged. Iveco R/2025/195 lists no number.
• These are mostly cars, vans, trucks and buses, with 29 motorcycle recalls and a wheelchair lift among the rest. Machines, panels and appliances see different stress and different reporting.
• The text is short and hedged. Most recalls say a fault "may" occur, and 229 of the 422 give no clue about when the fault got in.
• One fault can have several recall numbers. The 422 hold 385 distinct defect descriptions, as the same problem can be filed per brand or per model.
• It is no league table. Mercedes-Benz and Daimler account for 118 of the 422, and manufacturers differ in how much they file and how much detail they give.
• A word search still misses things. A recall about wiring that uses none of the search words is not here.

Download all 422 recalls with their labels (CSV, about 230 KB). Look up any recall number in the file to read the original words. For how a harness gets built in the first place, start with the guide to wire harness manufacturing.

Notes

1. Driver and Vehicle Standards Agency, Check vehicle recalls, recall data file RecallsFile.csv, and its Recalls Data User Guide, downloaded 6 November 2026. All recall numbers, vehicle counts and quotations here come from that file. Contains public sector information licensed under the Open Government Licence v3.0. ↩