Compliance 📅 October 4, 2026 ⏱ 15 min read

ETSI EN 302 208 Explained: Channels, Power, Beamwidth and Timing, With Clause Numbers

Channel plan of the EN 302 208 lower band, showing high-power interrogator channels 4, 7, 10 and 13 spaced 600 kHz apart with low-power tag-reply channels between them

EN 302 208 is the harmonised standard governing how a UHF RFID interrogator may use spectrum in Europe. It is the document a test lab, a notified body and a tender evaluator all reach for — and the one most often quoted from memory, which is how two pieces of folklore outlive the text.

What follows is the clause list that actually gets argued about, every figure quoted from the standard, ERC Recommendation 70-03 or the Official Journal, so you can link the clause rather than a 71-page PDF. We build ETSI-band and FCC-band reader variants configured per order, so this is the list our ETSI-band variants are configured against.

Two things worth knowing before the clause list

What replaced listen-before-talk in 2011

Since ETSI EN 302 208-1 V1.4.1 (2011-11) the standard permits multiple interrogators to transmit simultaneously on the same channel, and the foreword records that as the reason LBT ceased to be a requirement. The revision introduced “the ability for multiple interrogators to transmit simultaneously on the same channel”, which “provides significant improvements in spectrum efficiency and system performance. As a consequence ‘Listen Before Talk’ is no longer a requirement.” V1.4.1 carried LBT forward as an optional feature in its normative Annex B.

At V3.2.1 the mitigation method left the text altogether. Annex G of the current version gives the reason: “Following the European Commission implementing Decision (EU) 2018/1538 of 11 October 2018 the related update of CEPT 70 03 the mitigation method has been removed.” A search of V3.3.1 and V3.4.1 for “listen before talk” or “LBT” returns zero hits in both. What the current texts specify in its place is the clause 3.1 definition of adaptive frequency agility — changing channel automatically — and of dense interrogator mode, where multiple interrogators transmit in the same channel while tags reply in the adjacent ones.

Two live versions: V3.4.1 to download, V3.3.1 in the Official Journal

V3.4.1 (2023-12) is the newest published version. Presumption of conformity under the Radio Equipment Directive rests with V3.3.1, so a procurement document that names one version is making a choice between the newer text and the cited one. Both choices are defensible; naming the version is what matters.

Scope, bands and the version the Official Journal cites

The cover title of V3.4.1 (2023-12) is Radio Frequency Identification Equipment operating in the band 865 MHz to 868 MHz with power levels up to 2 W and in the band 915 MHz to 921 MHz with power levels up to 4 W; Harmonised Standard for access to radio spectrum. It runs to 71 pages; the national transposition dates printed on its front pages are adoption 18 December 2023, doa 31 March 2024, dop/e 30 September 2024, dow 30 September 2025.

Presumption of conformity is conferred by the Commission’s summary list of harmonised standards under Directive 2014/53/EU, which carries exactly two rows for this standard. The copy used here is stamped “Generated 7 September 2026” on its first line.

Version	Start of legal effect	Publication	End of legal effect	Withdrawal
V3.1.1	12.04.2017	OJ C 118, 12.04.2017	20.01.2023	OJ L 258, Decision 2021/1196, 20.07.2021
V3.3.1	20.07.2021	OJ L 258, Decision 2021/1196, 20.07.2021	open-ended	—
V3.4.1	Listed: V3.1.1 and V3.3.1 only.

The V3.3.1 row also carries a correction notice, below. For a buyer this reduces to one RFP question: which version does the test report name? A report against V3.3.1 maps onto the Official Journal row; one against V3.4.1 is against the newer, cleaner text. Ask for the version explicitly rather than for “EN 302 208” unqualified, and re-check the list the day you evaluate, because it is regenerated.

The channel plan, and why the channels are numbered 4, 7, 10 and 13

Why are Europe’s four high-power channels numbered 4, 7, 10 and 13 rather than 1 to 4? The numbers belong to a 200 kHz raster across the whole 865–868 MHz band, and every third slot is a high-power interrogator channel.

ERC Recommendation 70-03, Edition of February 2025, gives the formula: channel centre frequencies are 864.9 MHz + (0.2 MHz × channel number), with sub-band b2 using channel numbers 4 to 13. That text sits in Annex B, the informative annex covering references to legacy bands, which is why the numbering looks orphaned: the raster moved there on 14 February 2025 while EN 302 208 Table 1 kept the numbers. Running it, n = 4 gives 865.7, n = 7 gives 866.3, n = 10 gives 866.9 and n = 13 gives 867.5 MHz.

Clause 4.2.2.1 states it from the other direction: interrogators in the lower band “shall use any of the four specified high power channels”, the lowest channel centre is 865.7 MHz, each high-power channel is 200 kHz wide, the remaining three are spaced at equal intervals of 600 kHz — three raster steps — and “Tags shall respond within the low power channels.” So channel 4 spans 865.6–865.8 MHz and channel 13 spans 867.4–867.6 MHz, which is why the usable span is quoted as 865.6–867.6 MHz.

ERC 70-03 prints that span directly: Annex B sub-band b2 is listed as 865.6–867.6 MHz at 2 W e.r.p., bandwidth ≤ 200 kHz. Decision (EU) 2025/105 reaches the same place from the channel centres — its band entry 47a, the RFID entry, gives the band as 865–868 MHz with interrogator transmissions at 2 W e.r.p. “only permitted within the channels centred at 865,7 MHz, 866,3 MHz, 866,9 MHz and 867,5 MHz”, bandwidth ≤ 200 kHz. Entry 47b of the same annex is a separate row with its own limits: non-specific short-range devices for data networks at 500 mW e.r.p., with adaptive power control, duty cycles and the four 200 kHz ranges printed individually.

Ch	Centre	Sub-band	Role
1–3	865.1, 865.3, 865.5 MHz	b1	Low power / tag reply
4	865.7 MHz	b2	Interrogator (2 W)
5–6	865.9, 866.1 MHz	b2	Low power / tag reply
7	866.3 MHz	b2	Interrogator (2 W)
8–9	866.5, 866.7 MHz	b2	Low power / tag reply
10	866.9 MHz	b2	Interrogator (2 W)
11–12	867.1, 867.3 MHz	b2	Low power / tag reply
13	867.5 MHz	b2	Interrogator (2 W)
14–15	867.7, 867.9 MHz	b3	Low power / tag reply

The reply column is deliberately broad. Clause 4.2.2.1 says only that “Tags shall respond within the low power channels”, and footnote [6] of Decision (EU) 2025/105 places the reply “in a frequency range around the RFID interrogator channels”, so every low-power channel — b1, b3 and the b2 slots between the high-power channels — is available for it.

The upper band repays a careful read. Clause 4.2.2.2 says interrogators “shall use any of the three specified high power channels”, 400 kHz wide and spaced at 1.2 MHz — while Table 2 beneath it lists four rows: channels 3, 6, 9 and 12 at 916.3, 917.5, 918.7 and 919.9 MHz. The fourth carries a note that, per ERC 70-03 Annex 11, “there are existing implementations in some countries including a fourth RFID interrogator channel at centre frequency 919,9 MHz”. If a specification says three channels and a datasheet says four, this is why both are right.

Power, and the beamwidth ladder that sets it

Clause 4.3.3.3.1 sets the lower-band ceiling: e.r.p. on each of the four high-power channels “shall not exceed 2 W e.r.p. (33 dBm e.r.p.) specified in a bandwidth of 200 kHz”. Clause 4.3.3.3.2 sets the upper band at 4 W e.r.p. (36 dBm e.r.p.) in 400 kHz.

What is routinely missed: the 2 W step is earned by beam-width. Clause 4.3.4.2 defines beam-width as “the angle between the two half-power (-3 dB) points of the main lobe, when referenced to the peak effective radiated power of the main lobe”, and clause 4.3.4.3 grants the top rung to antennas of 90° or narrower, in three steps per band.

Lower band (865–868 MHz)	Beam-width	Upper band (915–921 MHz)	Beam-width
≤ 500 mW e.r.p.	Unrestricted	≤ 1 000 mW e.r.p.	Unrestricted
> 500 mW to ≤ 1 000 mW	≤ 180°	> 1 000 mW to ≤ 2 000 mW	≤ 180°
> 1 000 mW to 2 000 mW	≤ 90°	> 2 000 mW to 4 000 mW	≤ 90°

ERC Recommendation 70-03 reinforces this independently: Annex 11, technical parameters for sub-band a), “In addition, antenna beamwidth limits shall be observed as described in the standard EN 302 208”.

Worked against real panels, the ladder shapes procurement. A 105° panel sits on the middle rung, legal to 1 000 mW e.r.p. — 30 dBm, where 2 000 mW is 33 dBm, a 3 dB step. A 70° or 80° panel is inside 90° and reaches the full 2 000 mW rung. The wider antenna therefore carries two compounding effects, its lower gain and the lower rung it sits on. Specify beam-width and power together, and a 2 W-rated reader delivers the full 33 dBm e.r.p. step with a 70° or 80° panel in front of it.

The 4 s / 100 ms rule is a channel-release rule, not a duty cycle

This clause is quoted more often than it is read, usually as though it were a duty cycle. Clause 4.3.7.3 sets two limits on repeated transmissions on the same channel: “the on-duration of A shall not exceed 4 s” and “the off-duration of B shall be not less than 100 ms”. An interrogator may switch between channels at intervals not exceeding 4 s, and “shall not return to a previous channel within a period of less than 100 ms”. The upper band is specified without a transmission-length limit: the standard states that “There is no specific limit to the length of transmission … in the upper band”.

Worst case on one channel is 4 s on, 100 ms off:

4 / (4 + 0.1) = 0.9756 → 97.56 % maximum occupancy of a single channel.

Take a four-channel round robin at 4 s per channel: before returning to a channel the reader visits the other three, so it is away 3 × 4 s = 12 s — against a 100 ms minimum, 120 times the required gap. A hopping reader clears that gap by two orders of magnitude.

The binding obligation is the softer sentence preceding it: the measured transmission “shall be no greater than is required to read the tags present in the field and to verify that there are no additional tags present”, realized for example “by means of triggers for motion, light beam or by applying a duty cycles for polling.” Clause 4.2.1 adds a presence-sensing mode on the same logic: each transmission under 1 s, at least 100 ms between. A photo-eye trigger on a dock door is therefore a compliance feature as much as an efficiency one. The clause 4.3.7.2 NOTE gives the reason — “in order to ensure most efficient use of available channels for the general benefit of all users” — and Annex A, Table A.1 row 8 scopes it: it “Applies to interrogators in the lower band”.

For exporters, India tracks the same clause. The Short Range Devices (Exemption from Licence) Rules, 2021 — G.S.R. 853(E) of 10 December 2021, made expressly in supersession of the 2005 rules — reproduce the 4 s / 100 ms limits in Table-IV together with the standard’s own rationale, “in order to ensure most efficient use of available channels for the general benefit of all users”, name EN 302 208 in the standards column, and carry the same four channels centred at 865.7, 866.3, 866.9 and 867.5 MHz, the same 2 W e.r.p. and the same −20 dBm e.r.p. tag reply.

Interrogator categories and the receiver floors a datasheet must beat

EN 302 208 sets three receiver specifications, selected by transmit power, so the floor your datasheet must beat follows from the power you declare. Clause 4.2.4: an interrogator not exceeding 13 dBm e.r.p. is Category III; exceeding 13 dBm but not exceeding 30 dBm is Category II; all others are Category I. A 2 W portal reader is 33 dBm e.r.p., so it is Category I.

Table 4 sets the limits, and clause 4.4.5.3 is explicit that “the limit in table 4 is for the interrogator receiver without antenna” — worth settling before any datasheet comparison, since a figure that includes antenna gain is a different quantity.

Category	e.r.p.	Receiver sensitivity (Table 4)
Category I	> 30 dBm e.r.p.	−60 dBm
Category II	> 13 to 30 dBm e.r.p.	−55 dBm
Category III	≤ 13 dBm e.r.p.	−45 dBm

The rest applies to all interrogators regardless of category:

Adjacent channel selectivity (4.4.1): ≥ −26 dBm, adjacent channel at ±0.6 MHz lower band, ±1.2 MHz upper.
Blocking or desensitization (4.4.2.3): ≥ −23 dBm at fc ±2 MHz, ≥ −14 dBm at ±5 MHz, ≥ −8 dBm at ±10 MHz.
Spurious response rejection (4.4.4.3): ≥ −34 dBm at fc/3 and fc/2.
RF intermodulation (4.4.6.3): ≥ −34 dBm, f1 and f2 at fc + 0.6 MHz and fc + 1.2 MHz in the lower band.
Transmitter frequency error (4.3.1.3): shall not exceed ±10 ppm.

The plain-English gloss lives in the older text. V3.3.1 carried a descriptive Table 1b that V3.4.1 replaced with clause body text, describing Category I as the level used where tag distance is “often greater than 10 m or in challenging RF propagation environments”, with a dock door of a warehouse as its example — the warehouse and pallet-tracking case. V3.4.1 tightened the definition itself: under V3.3.1’s wording a 10 dBm reader satisfied both the Category II and the Category III sentences, and V3.4.1 resolves that. Read V3.3.1 for descriptions, V3.4.1 for definitions.

What the tag is allowed to send back

Clause 4.5.1.3 caps the tag: in the lower band radiated power “shall not exceed −20 dBm e.r.p. which is equivalent to a power spectrum density of −25 dBm/100 kHz e.r.p.”, and in the upper band −10 dBm e.r.p., equivalent to −18 dBm/100 kHz.

The standard shows its derivation in a NOTE, worth re-running because the printed values are rounded. The lower band assumes a 320 kHz tag emission bandwidth, so referencing to 100 kHz costs 10 log₁₀(3.2) = 5.05 dB, giving −20 − 5.05 = −25.05 dBm/100 kHz, printed as −25. The upper band assumes 640 kHz: 10 log₁₀(6.4) = 8.06 dB, giving −18.06 dBm/100 kHz, printed as −18. The standard checks itself backwards with “Prove: 5 dB = 10log10(k) k~3,2” and “8 dB … k~6,4”; strictly 10^(5/10) is 3.162 and 10^(8/10) is 6.310, so those are the bandwidth ratios restated to one decimal place.

Two boundaries define the tag’s spectral footprint. Clause 4.5.2.3 puts the tag emission mask edges at fc ±400 kHz lower band and fc ±800 kHz upper. Clause 4.3.6.2 sets where the interrogator’s spurious domain begins, following ERC Recommendation 74-01: beyond ±250 % of the channel separation, “which means beyond ±500 kHz for the lower band and ±1 000 kHz for the upper band”.

The consequence for range is structural. The forward link is capped at 2 W e.r.p. and the return link by law at −20 dBm e.r.p., so once you are at the ceiling the remaining levers are receiver sensitivity, antenna gain within the beam-width ladder, and tag placement — which is why the Category I −60 dBm floor matters more to real read range than the headline watts.

Spurious emission limits, and the correction the Commission published

Clause 4.3.6.3 points at Table 3, short enough to reproduce in full — e.r.p. limits, following ERC Recommendation 74-01.

State	87.5–118, 174–230, 470–694 MHz	Other frequencies below 1 000 MHz	Above 1 000 MHz
Operating	4 nW (−54 dBm)	250 nW (−36 dBm)	1 µW (−30 dBm)
Standby	2 nW (−57 dBm)	2 nW (−57 dBm)	20 nW (−47 dBm)
Resolution bandwidth	100 kHz	100 kHz	1 MHz

The top of that first column is where the correction lives. V3.3.1 printed the band as 470 MHz to 692 MHz. The Commission’s summary list carries a notice against the V3.3.1 row, in full: “For the purposes of presumption of conformity with the essential requirement set out in Article 3(2) of Directive 2014/53/EU, in Table 2 of this harmonised standard, the limit ‘692 MHz’ is replaced by the following: ‘694 MHz.’.” V3.4.1 incorporates the fix, renumbers the table 2 to 3 and prints 470 MHz to 694 MHz; Annex G records the revision as “Editorial corrections. Error correction in table 3. Replacement of figure 6 with correct figure.”

The effect is narrow and real. 694 MHz is the upper edge of the European broadcast band, exactly where mobile and broadcast services need protection. A lab working from an uncorrected V3.3.1 would treat 692–694 MHz as “other frequencies below 1 000 MHz” and apply 250 nW in place of 4 nW — a 2 MHz slice assessed about 18 dB too loosely (250 / 4 = 62.5, or 17.96 dB), exactly where mobile and broadcast services need protection. If you read V3.3.1 because it is the OJ-cited version, read the notice with it.

Version history, and how to cite it in a tender

The History page explains why older citations look structurally different. Most versions up to V2.1.1 were two-part deliverables; V1.1.2 of July 2006 was published as part 1 alone, and from V3.1.1 the standard is a single document.

Version	Date	Form
V1.1.1	September 2004	Parts 1 and 2
V1.1.2	July 2006	Part 1
V1.2.1	April 2008	Parts 1 and 2
V1.3.1	February 2010	Parts 1 and 2
V1.4.1	November 2011	Parts 1 and 2; LBT optional
V2.1.1	February 2015	Parts 1 and 2
V3.1.1	November 2016	Single document; OJ-cited 12.04.2017
V3.3.1	August 2020	75 pages; OJ-cited, current
V3.4.1	December 2023	71 pages; newest published

Two further revisions live only in the change log: Annex G records V3.1.2 and V3.2.1, and the History page records V3.4.0 of June 2022 at EN Approval Procedure stage. So although the mitigation method went at V3.2.1, the first published version carrying that structure is V3.3.1. The LBT timeline in full: a requirement before 2011; optional at V1.4.1, whose normative Annex B recommended a listen period of not less than 5 ms (clause B.1.2.1) and set the listen-mode threshold at “−35 dBm e.r.p. … or less depending on the nature of the application” (clause B.2.3); removed at V3.2.1; and from V3.3.1 onward the texts specify adaptive frequency agility and dense interrogator mode in its place.

Table numbers moved, so cite the version

V3.3.1’s Table 1a (upper-band frequencies) became V3.4.1’s Table 2; Table 1b (interrogator categories) dissolved into clause 4.2.4 body text; Table 2 (spurious emissions) became Table 3; Table 2a (receiver sensitivity) became Table 4. So “Table 2” means spurious emissions in V3.3.1 and upper-band channels in V3.4.1 — which is also why the Commission’s notice says “Table 2” while the corrected table in V3.4.1 is Table 3.

Where the law sits

The standard is the route to presumption of conformity; the spectrum itself is designated by Commission decisions. For the lower band that is Decision 2006/771/EC as amended, most recently by Implementing Decision (EU) 2025/105 of 22 January 2025 (OJ L series, 2025/105, 23.1.2025), whose band entry 47a permits 2 W e.r.p. “only permitted within the channels centred at 865,7 MHz, 866,3 MHz, 866,9 MHz and 867,5 MHz”, bandwidth ≤ 200 kHz. For the upper band it is Implementing Decision (EU) 2018/1538 of 11 October 2018 (OJ L 257/57, 15.10.2018), as amended by (EU) 2022/172 of 7 February 2022 (OJ L 28/21, 9.2.2022).

Footnote [7] of the 2025/105 annex ties the layers together and answers the LBT question at law: techniques to access spectrum and mitigate interference “shall be used”, and where relevant techniques “are described in harmonised standards or parts thereof the references of which have been published in the Official Journal of the European Union under Directive 2014/53/EU, performance at least equivalent to these techniques shall be ensured.” The law asks for equivalent performance; EN 302 208 defines what that is, today the channel-release discipline plus dense interrogator mode.

For a tender clause, name the standard, version, band and category — for example “EN 302 208 V3.3.1, lower band 865.6–867.6 MHz, Category I, 2 W e.r.p., antenna beam-width ≤ 90°” — and ask for the test report that matches it. Across several regions, our notes on exporting reader hardware and managing deployed reader fleets cover the configuration side.

Frequently asked questions

Does EN 302 208 require listen before talk?

Adaptive frequency agility and dense interrogator mode carry that function today. LBT ceased to be a requirement at EN 302 208-1 V1.4.1 (2011-11), whose foreword gives the reason: the revision introduced the ability for multiple interrogators to transmit simultaneously on the same channel, and “As a consequence ‘Listen Before Talk’ is no longer a requirement.” V1.4.1 carried it on as an optional feature in its normative Annex B. Annex G of the current text records that the mitigation method was removed at V3.2.1 following Decision (EU) 2018/1538, and a search of V3.3.1 and V3.4.1 for the term returns zero hits in both.

Which version of EN 302 208 gives presumption of conformity under the RED?

V3.3.1. The Commission’s summary list of harmonised standards under Directive 2014/53/EU carries two rows: V3.1.1, whose legal effect ran from 12.04.2017 and ended 20.01.2023, withdrawn by Decision 2021/1196 (OJ L 258, 20.07.2021); and V3.3.1, in legal effect from 20.07.2021 with an open-ended entry. V3.4.1 (2023-12) is the newer and cleaner text, and the list names V3.1.1 and V3.3.1. Ask which version a test report names rather than accepting “EN 302 208” unqualified.

Why are the EN 302 208 channels numbered 4, 7, 10 and 13?

Because the numbers index a 200 kHz raster across the whole band, so they count every slot rather than the high-power channels alone. ERC Recommendation 70-03 gives channel centre frequencies as 864.9 MHz + (0.2 MHz × channel number). So n = 4 gives 865.7 MHz, n = 7 gives 866.3 MHz, n = 10 gives 866.9 MHz and n = 13 gives 867.5 MHz — every third slot, matching clause 4.2.2.1’s 600 kHz spacing. The low-power channels around them are where tags reply.

Can I run 2 W e.r.p. into an omnidirectional antenna in Europe?

Clause 4.3.4.3 ties power to beam-width in three steps. An omnidirectional pattern sits on the unrestricted rung, which runs to 500 mW e.r.p.; above 500 mW to 1 000 mW the limit is 180° or narrower; and the full 2 000 mW step is granted to antennas of 90° or narrower. A directional panel of 70° or 80° is therefore the route to the 2 W ceiling, while a 105° panel reaches 1 000 mW. Specify beam-width and power together.

Is the 4 second / 100 ms rule a duty cycle limit?

It is a channel-release rule. Clause 4.3.7.3 caps continuous transmission on one channel at 4 s and requires at least 100 ms off before returning to that channel. That permits 4 / (4 + 0.1) = 97.56 % occupancy of a single channel, and a four-channel round robin returns after 12 s, 120 times the minimum gap. The binding obligation is the softer one: transmit only as long as needed to read the tags present. Per Annex A, Table A.1 row 8 it applies to interrogators in the lower band.

What receiver sensitivity does EN 302 208 require from a 2 W reader?

−60 dBm. A 2 W reader is 33 dBm e.r.p., and clause 4.2.4 makes anything above 30 dBm e.r.p. a Category I interrogator. Table 4 sets Category I at −60 dBm, Category II at −55 dBm and Category III at −45 dBm. Clause 4.4.5.3 adds that the limit is for the interrogator receiver without antenna, so a datasheet figure that folds in antenna gain is measuring a different quantity.

How much power is an RFID tag allowed to reply with?

Clause 4.5.1.3 caps the tag at −20 dBm e.r.p. in the lower band, equivalent to −25 dBm/100 kHz, and −10 dBm e.r.p. in the upper band, equivalent to −18 dBm/100 kHz. The lower-band figure assumes a 320 kHz emission bandwidth: 10 log₁₀(3.2) = 5.05 dB, so −20 − 5.05 = −25.05 dBm/100 kHz, printed as −25. Because the return path is capped by law, receiver sensitivity and antenna gain carry the link budget rather than transmit power.