What Size Cable Do I Need? A UK Guide to BS 7671

BS 7671 sizes cable by making four currents agree with each other. They have unhelpful names and there's no way round learning them.

Ib is the design current. What the circuit will genuinely draw. For a 9.5kW shower on 230V, that's 9,500 divided by 230, so 41.3A.

In is the rating of the breaker or fuse protecting the circuit. It has to be at least Ib, and it has to be a size that actually exists. UK standard ratings run 6, 10, 16, 20, 25, 32, 40, 50, 63A and upwards. For our 41.3A shower, the next real size up is 50A.

It is the minimum current-carrying capacity the cable needs to show in the tables, before anything gets derated.

Iz is the honest number. What the cable can actually carry, sitting where you've actually put it, after the correction factors have had their say.

The whole method is one rule: Ib is less than or equal to In, which is less than or equal to Iz. If that chain holds, and voltage drop is inside limits, the cable is right.

Takeaway: if someone quotes you a cable size without asking how it's installed and how long the run is, they've answered a different question.

Here's the part that gets skipped, and it costs more than any other input.

BS 7671 publishes current-carrying capacity against an installation reference method. Same cable, different surroundings, different number. A 2.5mm² twin and earth clipped direct to a joist carries 27A. Take that identical cable and bury it in loft insulation without touching the wall, which the regulations call Method 103, and it carries 13.5A.

Half. Same copper, same insulation, same everything except where it sits.

The reason is heat. Cable capacity is a thermal limit, not an electrical one. The conductor is rated to run at 70°C for standard thermoplastic twin and earth. Clipped to a joist in open air, heat sheds into the room. Packed in 200mm of loft insulation, it can't go anywhere, so the same current cooks the cable to a temperature the insulation was never designed for.

If a run passes through different conditions, and most do, size the whole circuit for the worst section. A cable that's clipped direct for 15 metres and buried in insulation for two is a Method 103 cable.

Takeaway: the installation method is the single input most worth getting right, and Method 103 is the one that bites.

Once you've got a tabulated number, four factors adjust it for reality. Multiply them together to get Ct, then divide the device rating by Ct to find the minimum capacity you need from the tables.

Ca, ambient temperature. The tables assume 30°C. A cable in a plant room at 45°C loses about 21% of its capacity. A cold loft in February gains a little back, though nobody designs for that.

Cg, grouping. Cables bunched together heat each other. Two circuits in a conduit and each one derates to 80%. Six circuits and you're at 57%. This is why a stuffed conduit is worth avoiding.

Ci, thermal insulation. A cable totally surrounded by insulation over a length of half a metre or more takes a 0.5 factor under Regulation 523.9.

Cc, device type. Only bites if there's a rewireable fuse to BS 3036 in the circuit, which carries a 0.725 factor because of how loosely it blows.

One trap worth naming. If you've already selected Method 100 to 103, the insulation is baked into the tabulated number. Applying Ci as well double-counts it and you'll oversize the cable. Our cable sizing calculator blocks that combination for exactly this reason.

Takeaway: correction factors turn a table value into a real one, and the insulation factor is the one people apply twice.

Twin and earth, clipped direct along joists, one circuit on its own, 30°C loft, Type B MCB.

Design current: 9,500 / 230 = 41.3A

Device: next standard size up is 50A

Correction factors: Ca 1.00, Cg 1.00, Ci 1.00, Cc 1.00. Combined, Ct = 1.00

Minimum capacity needed: 50 / 1.00 = 50A

Cable: 6mm² clipped direct carries 47A, which is short. 10mm² carries 64A. So 10mm².

Voltage drop: 10mm² drops 4.4 mV/A/m. Over 18 metres at 41.3A that's (4.4 × 41.3 × 18) / 1000 = 3.27V, or 1.42% of 230V. The limit is 5%. Passes easily.

Answer: 10mm² twin and earth on a 50A Type B MCB, with a 4mm² CPC. Which is what you'd pull off the shelf at CEF or Screwfix for a shower that size anyway, because the sums land where the trade practice already sits.

Takeaway: on short shower runs the current decides everything and voltage drop never gets a vote.

Now the same cable type, a smaller load, and a much longer run. Twin and earth, clipped direct, Type B MCB.

Design current: 6,000 / 230 = 26.1A

Device: 32A

Cable on capacity: 4mm² clipped direct carries 37A, comfortably over 32A. Done, apparently.

Voltage drop: 4mm² drops 11 mV/A/m. Over 45 metres at 26.1A that's (11 × 26.1 × 45) / 1000 = 12.9V. On 230V that's 5.6%. The limit is 5%. It fails.

Step up to 6mm²: (7.3 × 26.1 × 45) / 1000 = 8.57V, which is 3.7%. That passes.

Answer: 6mm², not 4mm². The load didn't change. The distance did. This is the example worth remembering, because a 4mm² garage supply looks completely reasonable right up until someone measures the voltage at the far end.

Takeaway: past about 30 metres on a domestic circuit, assume voltage drop is in charge until you've proved it isn't.

BS 7671 Appendix 4 gives 3% for lighting and 5% for everything else, measured from the origin of the installation to the point of use. On 230V that's 6.9V and 11.5V.

Now the bit that surprises people who've been quoting those numbers for years. Appendix 4 is informative. It's guidance, not a regulation. The actual requirement is Regulation 525.202, which says the voltage at the terminals of fixed current-using equipment has to be no less than the standard for that equipment.

In practice this changes nothing for most work. The 3% and 5% figures are how you demonstrate compliance, every UK inspector expects to see them, and arguing the point on site will win you nothing. But when there's sensitive kit at the end of a long run, and the manufacturer specifies a tighter tolerance than 5%, their figure is the one that governs. The regulation cares about the equipment, not the percentage.

Takeaway: use 3% and 5% as working limits, and read the equipment data sheet when anything fussy is on the end.

Rules of thumb only — each assumes a sensible installation method and a run under about 25 metres. In practice most UK sparks size a ring final at 2.5mm² and never think about it again, and they're right to, because a ring in a normal house is nowhere near its limits. The circuits that actually need working out are the long ones and the hot ones: garages, outbuildings, EV chargers on a detached drive, anything crossing a loft stuffed with insulation.

The method isn't complicated, it's just fiddly, and there are four places to make an arithmetic slip. Our cable sizing calculator runs the full BS 7671 sequence: design current, device selection, correction factors, capacity check and voltage drop check, and it tells you when voltage drop rather than current is what forced the cable size. If you only want the volt drop on a cable you've already picked, the voltage drop calculator handles that without the full sizing method.

Both are free and neither is a substitute for a design by a competent person.

Cable sizing is two checks out of a much longer list. A full BS 7671 design also covers earth fault loop impedance, prospective fault current, disconnection times, discrimination between protective devices, diversity across the installation, and thermal withstand under fault conditions. This guide and the calculator handle current-carrying capacity and voltage drop, which is enough to specify a cable and nowhere near enough to sign off an installation.

Check every figure against the current edition of BS 7671 and the IET On-Site Guide before you rely on it. Where a manufacturer publishes data for their specific cable, Doncaster Cables and Prysmian both do, that data beats a generic table.

Takeaway: the sums get you a cable size. A competent person gets you a safe installation.