Insulation Resistance Testing: Voltages, Minimums and Method

Insulation resistance (IR) testing puts a d.c. voltage across conductors that should be insulated from each other and measures the leakage in megohms. Table 64 of BS 7671:2018+A4:2026 fixes the test voltage and the pass mark: 250 V d.c. and 0.5 MΩ for SELV and PELV, 500 V d.c. and 1.0 MΩ for anything up to and including 500 V (every normal 230 V and 400 V circuit), and 1000 V d.c. and 1.0 MΩ above that. It is a dead test done after continuity and before anything is made live, and Regulation 643.3 says how to do it, when line and neutral may be joined, and when 250 V is allowed. This guide works through the regulation, the method on new work and on EICRs, and what the numbers mean.

What does an insulation resistance test measure?

The instrument applies a steady d.c. voltage between two conductors and measures the small current that leaks through the insulation separating them. Resistance is voltage divided by current, displayed in MΩ. Sound PVC insulation on a short domestic circuit leaks so little that the display goes over range. Insulation that is cracked, crushed, overheated, damp or contaminated leaks more and the figure drops.

The test voltage is deliberately higher than the circuit's working voltage, so weaknesses that would only show under stress are found on the bench rather than in service. That is also why the voltage can damage things: 500 V d.c. across the terminals of a dimmer or a USB socket is not something the manufacturer designed for.

Regulation 643.3.1 says the measurement is made between live conductors, and between live conductors and the protective conductor connected to the earthing arrangement, and adds that for this measurement line and neutral may be connected together. In practice that gives you three possible readings on a single-phase circuit (L-N, L-E, N-E) or the combined L+N to E, and on three-phase the live-to-live combinations L1-L2, L2-L3, L1-L3 and each line to neutral, plus each live conductor (or all of them joined) to earth.

What test voltage do I use and what is the minimum reading?

These are the Table 64 figures, and they are minimum values below which the circuit fails. They are not targets. A new circuit will read hundreds of megohms or over range; an older installation in service is usually tens of megohms or more. IET Guidance Note 3 (now in its 10th Edition for A4) adds the figure most electricians work to in practice: where an insulation resistance below 2 MΩ is recorded, the possibility of a latent defect exists, and each circuit should be tested separately to find where the leakage is.

Regulation 643.3.2 also says what you are testing: the main switchboard and each distribution circuit tested separately, with all its final circuits connected but with current-using equipment disconnected, must meet the Table 64 value. The generic schedule of test results asks for a figure per circuit, and that is what assessors expect, but the regulation itself is written around the assembled installation. Both matter, which is why the whole-board-then-split routine in the next section works.

How do SPDs and electronics change the test?

A surge protective device is designed to conduct when the voltage across it goes above its maximum continuous operating voltage, Uc, and for a 230 V device that is a little above the 253 V upper tolerance of the supply. Put 500 V d.c. across it and it does exactly what it is designed to do, which is why NICEIC's guidance (10 September 2021) explains that an SPD left connected will drag the reading down and may be damaged.

Regulation 643.3.2 gives two options. Where SPDs or other equipment are likely to influence the test or be damaged, disconnect them before testing. Where that is not reasonably practicable, for example a socket-outlet with a built-in SPD, the test voltage for that circuit may be reduced to 250 V d.c., but the insulation resistance must still be at least 1 MΩ. NICEIC's advice is to disconnect wherever possible and, crucially, to reconnect every SPD afterwards; a board SPD left on its "disconnected" setting or with the line link pulled gives the customer no protection and nobody will notice until a surge arrives.

Regulation 643.3.3 covers electronic devices in the circuit that could affect the result or be damaged. Since Amendment 2:2022 it reads as a two-stage requirement on new work: disconnect the electronics and test the wiring at the Table 64 voltage, then, following connection of the equipment, apply a test at 250 V d.c. between the live conductors and the protective conductor connected to the earthing arrangement, again needing at least 1 MΩ. The second test is the one that catches a cable pinched under a cover plate or a stray strand at the last accessory you fitted. GN3's reading of it is a single L+N to E test, which ensures every conductor is covered without putting a voltage across the equipment.

For the devices themselves, the usual list to lift, unplug or switch out of circuit is: dimmers, electronic timers and PIRs, USB socket-outlets, smoke and heat alarms (unclip from their bases), boiler and heating controls, LED drivers and fluorescent control gear, shaver sockets, extractor fans with timers, EV chargers, PV inverters, and the SPD at the board. For more on how SPDs are specified and when they are required, see Surge Protection: SPD Requirements under Regulation 443.

What is the method on an EIC for new work?

On new work nothing is connected yet, so there is no excuse for skipping the live-to-live test. The sequence that satisfies 643.3 and gives a clean schedule is:

1. Isolate at the board, lock off, and prove dead with a GS38 voltage indicator and proving unit. Check the instrument's leads: open leads should read over range, shorted leads should read zero.
2. Confirm every lamp, driver, control and SPD is out of circuit. Lights off at the switch separates line and neutral at the lampholder; for two-way and intermediate circuits, test with the switches in each position so every strapper is included.
3. Test L-N (or every live-to-live pair on three-phase) at 500 V d.c. This is the test that finds a line-to-neutral short before the circuit is energised for the first time.
4. Test L-E and N-E, or L+N to E, at 500 V d.c.
5. Let the reading settle. Long runs and screened cable charge like a capacitor and the figure climbs for a second or two; record the steady value. The instrument discharges the circuit automatically when you release the test button, but give it a moment before touching anything.
6. Reconnect the equipment and carry out the 643.3.3 test at 250 V d.c., L+N to E, confirming at least 1 MΩ.
7. Record the readings in the live/live and live/earth columns of the schedule, with the test voltage, and put the SPD back exactly as you found it.

Insulation resistance sits third in the initial verification order in Regulation 643.1, after continuity of protective conductors and bonding and continuity of ring final circuit conductors, and before polarity, earth electrode resistance and all the live tests. It comes after continuity because an open cpc or a broken ring leg needs fixing before you stress the circuit, and before the live tests because nobody wants to energise a line-to-earth short. For the tests either side of it see Continuity Testing: the R1+R2 Method, Ring Final Circuit Testing: r1, rn and r2 and Polarity Testing Explained.

How do I test an installation in service for an EICR?

On a periodic inspection the installation is live, full of equipment and belongs to somebody else, so full disconnection of every item is rarely reasonably practicable. The accepted approach, described in GN3 and underpinned by the "may be connected together" wording of 643.3.1, is to join line and neutral and test between the combined live conductors and earth. Because L and N sit at the same potential during the test, nothing connected between them sees any voltage, so dimmers, drivers and alarms are safe. Anything connected between a live conductor and earth, such as a filter capacitor, a neon, a damp junction box or a failing heater element, still shows up.

The efficient routine is whole board first, then split. With the main switch open and the installation proved dead, every protective device closed and vulnerable items disconnected, test all the live conductors together to the main earthing terminal. If the figure is above 2 MΩ, GN3 regards the installation as sound and you can record per-circuit values by testing each circuit individually only where the schedule requires them, or test each circuit anyway if that is your scheme's expectation. If the whole-board figure is under 2 MΩ, subdivide: open the devices one at a time and watch the reading, and the circuit that lifts it is where the leakage is.

There is some arithmetic behind the 2 MΩ trigger. Insulation resistances in parallel combine like resistors in parallel, so the whole-installation figure is always lower than the weakest circuit. Ten circuits each at 20 MΩ give 2 MΩ at the board. A single circuit at 1.5 MΩ among nine healthy ones will pull the total below 1.5 MΩ. A low whole-board reading therefore does not mean ten faults, and a per-circuit reading just over 1 MΩ can be hiding behind a whole-board figure that looks fine only if the other circuits are over range.

On the schedule, record the L+N to E result in the live/earth column and enter LIM in the live/live column, with the limitation stated in Section D of the report: "Insulation resistance tested L+N to E with equipment connected; L-N not tested." That is a legitimate limitation, agreed with the person ordering the work, not a shortcut. The live-to-live test is less critical on a circuit that was energised that morning, because a line-to-neutral fault of any substance would have operated the protective device already. Our guide to Writing EICR Observations covers how limitations and observations are worded on the A4 model form.

Electronic RCDs and RCBOs can also influence the reading, since some contain components between neutral and earth or a functional earth connection. Where the manufacturer's instructions say so, test on the load side of the device or lift the neutral, and note what you did. If you are refreshing your RCD knowledge at the same time, see RCD Testing Explained.

How do I interpret a low reading?

A reading that collapses to zero on one circuit is far more often something still connected than a cable fault: a lamp left in, a boiler still wired, an immersion element, a neon in a switch. Eliminate that first. Genuine low readings come from a nail or screw through a cable, water in an outside socket or a loft junction box, a cable crushed behind a kitchen unit, rodent damage, perished rubber insulation on pre-1960s wiring, or a neutral touching the earth terminal inside an accessory.

To locate the fault, split the circuit. Disconnect at a junction or accessory roughly halfway and test each half; on a ring final, break the ring at the board and test each leg. Testing L-E and N-E separately, rather than linked, tells you which conductor is affected and narrows the search again. Low readings on every circuit at once, on a wet day in an old building, usually point to moisture in the board or a damp meter cupboard rather than ten faults, and often recover once the enclosure is opened and dried; re-test and note it. The step-by-step fault-finding process is in Low Insulation Resistance: Finding the Cause.

What code does low insulation resistance get on an EICR?

Electrical Safety First's Best Practice Guide 4, Issue 7.4 (August 2026), written against BS 7671:2018+A4:2026, lists one insulation resistance example: insulation resistance of less than 1 MΩ between live conductors connected together and Earth, coded C2, potentially dangerous. It gives no insulation resistance examples under C1, C3 or FI, and it says that for domestic and similar installations it should generally be possible to attribute a code to each observation without needing further investigation.

So a reading below 1 MΩ is a C2 and the report is unsatisfactory. A reading between 1 and 2 MΩ complies with Table 64; GN3 asks you to investigate, and the right outcome is to find what is causing the leakage and code that, whether it is a damp outside socket (C2 if there is exposed conductor, C3 or no code if it is a tired but intact enclosure) or a cable with a screw through it. Coding the number alone, without finding the cause, is where FI tends to creep in, and BPG4's position is that domestic work should rarely need it. NICEIC's older guidance on the point, from October 2018, suggested comparing the figure with the previous EIC or EICR: an unexpected fall in a few years is itself useful information for the recommended interval, even where the current figure passes. See EICR Observation Codes: C1, C2, C3 and FI for the full coding framework.

Remember that BPG4 also expects any C1, C2, C3 or FI in the schedule of test results to have a matching entry in the observations section, so a 0.4 MΩ figure in the live/earth column needs an observation with a location and a description of the condition, not just the number.

When do I record >999 MΩ or >200 MΩ?

Instruments have a top of range and show it with a ">" symbol when the real value is beyond it. The Megger MFT-X1 offers insulation test voltages of 50, 100, 250, 500 and 1000 V d.c. and measures from 0.01 MΩ to 999 MΩ, so its over-range display is >999 MΩ; many other testers top out at >200 MΩ or >1000 MΩ. Copy what the display shows onto the schedule, symbol included. The 50 V and 100 V ranges on the MFT-X1 are there for SELV and PELV extra-low-voltage circuits and sensitive equipment where even 250 V would be too much; they are not substitutes for the Table 64 voltage on a mains circuit. For those circuits see SELV, PELV and FELV Explained.

Do not write 999 or 200 on its own. Without the symbol it reads as a measured value, and an assessor who knows your instrument's specification will ask about it. Equally, do not round a genuine 85 MΩ up to over range. Insulation deteriorates slowly, and the only way the next inspector can see that a circuit has gone from 85 MΩ to 4 MΩ in five years is if both figures were recorded as measured. The Schedule of Test Results guide covers the rest of the columns.

Does the same test apply to a solar PV array?

The d.c. side of a PV system is tested to IEC 62446-1 rather than Table 64, and the figures are chosen by system voltage: 250 V d.c. with a 0.5 MΩ minimum for strings below 120 V, 500 V d.c. and 1 MΩ from 120 V to 500 V, and 1000 V d.c. and 1 MΩ above 500 V. Test between the array positive and negative (shorted together, or each separately) and earth, in daylight if the inverter and any functional earth are isolated, and record the test voltage and the irradiance. Section 712 of BS 7671 adds the a.c.-side requirements; see Solar PV Installation Certificate: Section 712.

How Certio helps

Certio produces the EICR, EIC, Minor Works certificate and PAT records on the BS 7671 model forms and turns them into branded PDFs. Photograph the consumer unit and the circuit schedule drafts itself; point the phone camera at a Megger MFT-X1 and the reading goes straight into the schedule of test results, or dictate it while your hands stay on the probes. Suggested observation wording with the regulation reference is there for you to check and confirm. The judgement stays yours; the typing goes. Starter is £4.99 a month, charged when you sign up.

Written by the Certio team. Last checked 16 October 2026.