IEC 60870-5-104 moves telecontrol traffic — the commands and status data between a control center and substation equipment — in cleartext over TCP/IP with no authentication built into the base protocol, a gap academic research has documented for over a decade. It is also, confirmed directly by ESET's and Mandiant's independent malware analyses, the only protocol Industroyer2 spoke to attack a Ukrainian energy provider's substation equipment in 2022, and Hitachi Energy RTU500 advisories through 2026 show implementation bugs in IEC-104 stacks are still being found and patched today.
Key takeaways
- The point for security is that none of these fields carries a key, a signature or a cryptographic check: the protocol trusts whoever holds the TCP session.
- Yes, split across two parts of the IEC 62351 security series — the same series that secures IEC 61850 and other TC 57 protocols (see IEC 62351) — plus a 60870-5 mapping document.
- Yes — this is one of the better-documented cases of a specific OT protocol being the actual attack mechanism, not just a theoretical risk.
- The exception is a test tool rather than a field device.
What is IEC 60870-5-104, and where does it run?
IEC 60870-5-104 is an IEC Technical Committee 57 standard, formally titled "Telecontrol equipment and systems – Part 5-104: Transmission protocols – Network access for IEC 60870-5-101 using standard transport profiles." It takes the older serial protocol IEC 60870-5-101 and carries it over TCP/IP instead, registered with IANA on TCP/UDP port 2404 as "IEC 60870-5-104 process control over IP." The current text is the consolidated Edition 2.1 (the 2006 second edition plus Amendment 1, with the August 2023 corrigendum included); IEC's webstore also lists a second amendment at stage ACDV with a forecast publication date of 3 April 2027. Industry practice widely treats it as the international and European counterpart to DNP3 — the protocol that fills the same role (control-center-to-substation telecontrol and monitoring) in North America. Where DNP3 findings dominate CISA's advisories for U.S. utilities, IEC-104 findings dominate the equivalent advisories for internationally deployed grid gear, most consistently in recent years against Hitachi Energy's RTU500 remote terminal unit line.
How is an IEC-104 message built from APCI and ASDU?
Every IEC-104 frame is an Application Protocol Data Unit (APDU): a short Application Protocol Control Information (APCI) header defined by 60870-5-104 itself, optionally followed by an Application Service Data Unit (ASDU) taken over from IEC 60870-5-101. The standard's contents page shows the parts a firmware team has to implement: the APCI definition, three control-field formats, loss and duplication protection, test procedures, start/stop transmission control, the port number and the "k" limit on unacknowledged I-format APDUs.
| Part | What it carries | Where parser bugs have appeared |
|---|---|---|
| Start octet and length | 0x68 followed by the APDU length (see the start/test frames in lib60870's server code) | Wrong length handling blocked the link layer in CVE-2023-5769 |
| I format | Numbered information transfer: send and receive sequence numbers plus an ASDU | Crafted I-frames caused heap over-reads in lib60870 |
| S format | Numbered supervisory acknowledgement, no ASDU | — |
| U format | Unnumbered control: STARTDT, STOPDT, TESTFR (activation and confirmation) | An invalid U-format frame caused denial of service in CVE-2026-1773 |
| ASDU header | Type identification, variable structure qualifier (object count, sequence bit), cause of transmission, originator and common address — the fields exposed by lib60870's ASDU API | An inflated object count drove both 2026 lib60870 CVEs |
| Information objects | Information Object Address (IOA) plus the value, for example a single or double command | Industroyer2 wrote commands straight to known IOAs |
The point for security is that none of these fields carries a key, a signature or a cryptographic check: the protocol trusts whoever holds the TCP session.
Does IEC-104 have any built-in security?
No. A 2014 IEEE PES paper studying intrusion detection for the protocol states directly: "the IEC 60870-5-104 protocol transmits messages in clear text without any authentication mechanism." A later survey attributes the protocol's core vulnerability class to exactly this gap, citing documented man-in-the-middle demonstrations enabled by "the lack of data integrity and authentication measures." NIST SP 800-82 Revision 3's general description of OT protocols captures the same pattern without naming IEC-104 specifically: "Many OT protocols have no authentication at any level. Without authentication, there is the potential to replay, modify, or spoof data or devices." In practice, on its standard port 2404, IEC-104 traffic is plain unencrypted TCP — readable and, absent any compensating control, injectable by anything with access to the same network segment.
Is there a secure version, and what does it protect?
Yes, split across two parts of the IEC 62351 security series — the same series that secures IEC 61850 and other TC 57 protocols (see IEC 62351) — plus a 60870-5 mapping document. It matters which generation a product implements, because IEC's own abstracts show the application-layer mechanism changed completely in 2023.
| Document | What it adds | Layer |
|---|---|---|
| IEC 62351-3 (Ed. 2.0, 2023) | TLS 1.2 and TLS 1.3 profile for TCP/IP protocols, including cipher suites, session resumption and renegotiation, CRL/OCSP revocation and security events; the referencing standard only fixes the port. IANA lists "iec-104-sec" on TCP 19998 | Transport (confidentiality + integrity) |
| IEC TS 62351-5:2013 with IEC TS 60870-5-7:2013 | Secure authentication with challenge/reply and "aggressive mode"; TS 60870-5-7:2013 defines the matching ASDUs, such as type 81 (authentication challenge) and type 83 (aggressive mode request) | Application |
| IEC 62351-5 (Ed. 1.0, 2023) | Per-association Station Association, Session Key Change and Secure Data exchange; challenge/reply removed, aggressive mode replaced, authenticated encryption added, sequence-number rules updated | Application (authenticity, optional confidentiality) |
| IEC TS 60870-5-7:2025 (Ed. 2.0) | Message formats for applying 62351-5:2023 and 62351-3:2023 to 101/104, plus role-based access control using IEC 62351-8 | Mapping for 101/104 |
The 62351-5:2023 preview explains why the standard brings its own counters: the existing protocol sequence numbers "have very low maximum values, permitting an attacker to attempt a replay after gathering only a small number of messages." It also names the four threats it addresses — spoofing, tampering, replay and eavesdropping — so denial of service stays out of its scope.
Vendor documentation shows these combined in deployed products: Siemens' SICAM/SIPROTEC hardening manual states that SICAM PAS "has an IEC 62351-3/-5 and IEC 60870-5-7 conform implementation via Security Add-on package" for DNP3 and IEC-104. The open-source lib60870 README says it "supports most TLS features required by IEC 62351-3" with mbedtls, and offers secure authentication to IEC 60870-5-7:2013 as a commercial add-on. Whether a specific RTU or control-center product has either layer enabled, and which edition it follows, is a question for that product's own documentation, not something to assume from the standard's existence. The DNP3 world went through the same split; see DNP3 Secure Authentication.
Has IEC-104 actually been used in a real attack?
Yes — this is one of the better-documented cases of a specific OT protocol being the actual attack mechanism, not just a theoretical risk. ESET's 2022 analysis of Industroyer2, the malware deployed against a Ukrainian energy provider, is explicit: "Industroyer2 only implements the IEC-104 (aka IEC 60870-5-104) protocol to communicate with industrial equipment. This includes protection relays, used in electrical substations." Mandiant's independent analysis of the same malware (which it tracks as INDUSTROYER.V2) describes the mechanism: the malware "crafts configurable IEC-104 Application Service Data Unit (ASDU) messages... to change the state of a remote station's Information Object Addresses (IOAs) to ON or OFF" — sending protocol-legitimate commands to toggle breaker states, something IEC-104's lack of authentication does nothing to block once an attacker has network access. Netresec's packet-level analysis of captured Industroyer2 traffic confirms the specific ASDU type IDs used (46 for double command, 45 for single command) against targeted breaker addresses, and notes each thread set an IOA "every 3 seconds no matter what the server responded" — the attackers already knew the station addresses and IOAs, so no enumeration traffic preceded the commands. The original 2016 Industroyer malware against the Ukrainian grid is widely reported, via ESET and Dragos research, to have carried a similar IEC-104 module among several protocol-specific payload modules; treat it as well-corroborated secondary reporting rather than a directly re-checked primary source.
What has CISA found in IEC-104 products recently?
Every CISA ICS advisory naming IEC-104 from 2023 through 2026 that this guide could verify affects Hitachi Energy's RTU500 product line, with one additional advisory against the open-source MZ Automation lib60870 stack and one against a protocol test set, described after the table. Every IEC-104 finding in the table is an availability or memory-safety flaw that crashes, restarts or blocks the unit; none is described as remote code execution or authentication bypass. CVE, CVSS v3 base score and precondition below are taken from each advisory's CSAF file:
| Advisory | IEC-104 CVE (CVSS v3) | Finding and precondition |
|---|---|---|
| ICSA-26-211-11 (lib60870 2.4.0) | CVE-2026-61893, CVE-2026-63033 (6.5 each) | One-byte heap over-read on a crafted I-frame with an inflated object count; fixed in 2.4.1 |
| ICSA-26-155-04 | CVE-2026-8479 (6.5) | "IEC 60870-5-104 used in bidirectional mode is vulnerable for a NULL pointer dereferencing"; only with bidirectional mode configured |
| ICSA-26-062-03 | CVE-2026-1773 (7.5) | Denial of service on an invalid U-format frame in bidirectional mode; 62351-3 mitigates but does not remediate |
| ICSA-25-093-01 | CVE-2024-11499 (4.9), CVE-2024-12169 (6.5) | Control-module restart: certificates updated while in use, or a specific sequence with 62351-3 TLS enabled |
| ICSA-25-128-02 | CVE-2023-5769 (5.9) | Link-layer blocking from wrong APDU length information |
| ICSA-24-354-01 | CVE-2023-6711 (5.9) | Buffer overflow in SCI/HCI IEC-104 components, forcing a reboot |
| ICSA-23-220-02 | CVE-2022-2502 (7.5), CVE-2022-4608 (7.5) | Stack-based overflows reachable only with 62351-5 "Advanced security" or with 62351-3 TLS configured |
The last rows are worth noting on their own: flaws inside the very features meant to secure IEC-104 are a reminder that enabling 62351-3 or 62351-5 adds its own implementation attack surface — it does not make the deployment immune to vulnerabilities, only changes what kind of flaw to look for.
The exception is a test tool rather than a field device. ICSA-26-239-04 (August 2026) covers Applied Systems Engineering's ASE2000 V2 communications test set: versions 2.35 through 2.37 do not properly validate certificates for IEC 60870-5-104 over TLS (CVE-2026-18717, 7.4), and versions 2.25 through 2.37 bundle an Apache log4net release older than 2.0.10 with an XML external entity flaw (CVE-2018-1285, 9.8). CISA says customers must upgrade to 2.38 or later for both fixes.
What should a firmware team shipping an IEC-104 stack do?
The advisories cluster in a small number of code areas. Mapping them to tests and evidence gives a short, defensible work list for an RTU, gateway or relay team:
| Code area | Test before release | Evidence to keep |
|---|---|---|
| APCI length and frame parsing | Fuzz length octet, truncated and delayed frames | Fuzz corpus and crash log per build |
| ASDU object count and IOA parsing | Fuzz type IDs with inflated object counts and sequence bit set | Same, plus sanitizer runs |
| U-format and session state machine | Invalid and out-of-order STARTDT/STOPDT/TESTFR, both directions | State-machine test report |
| TLS (62351-3) | Renegotiation, session resumption expiry, certificate update on a live session | TLS library name and version in the SBOM |
| Application security (62351-5) | Edition implemented (2013 TS or 2023) and key-change handling | Statement in the product security documentation |
| Third-party stack | Version tracking for lib60870 or a commercial stack | SBOM entry, licence record (lib60870 is GPLv3 or commercial) |
Two practical rules follow. First, list the protocol stack and the TLS library as separate components in a firmware SBOM, because both appear in advisories on their own. Second, publish advisories for your own product in a machine-readable form so utilities can match them; see CSAF and VEX for device makers.
Worked example: triaging an RTU build that embeds lib60870
Synthetic example, for illustration only: a hypothetical maker of a small feeder RTU.
- The release manager pulls the SBOM for firmware build 3.2 and finds lib60870 2.4.0 linked into the 104 server task, with mbedtls 2.28 for TLS.
- The GHSA records for lib60870 list 2.4.0 as vulnerable and 2.4.1 as patched for both 2026 over-reads (for example GHSA-7v97-jmwv-w5j7), and ICSA-26-211-11 scores each 6.5, network-reachable without authentication.
- The team checks exposure: the RTU accepts IEC-104 on port 2404 from the control center and has TLS on 19998 as an option. TLS does not remove the parser, so both ports reach the vulnerable code.
- The fix is a library bump to 2.4.1, then the fuzz and state-machine tests from the table above, then a regression run of the interoperability profile.
- The maker issues its own advisory naming the affected builds, the fixed build and the CVE IDs, and updates the SBOM for the new build.
What should an asset owner running IEC-104 equipment check first?
Inventory every RTU, protection relay and control-center gateway speaking IEC-104, with exact firmware versions — every advisory above affects specific version ranges, and the DoS findings against Hitachi Energy RTU500 recur across years, meaning an unpatched unit stays exposed across multiple disclosure cycles. Confirm whether IEC 62351-3 (TLS) and 62351-5 (authentication) are actually configured, not just supported in the product's documentation — and if either is enabled, confirm the firmware version is current, since both features have had remotely exploitable vulnerabilities of their own (ICSA-23-220-02). Segment IEC-104 traffic away from business networks and the internetThe protocol was designed for a trusted telecontrol network, not public exposure, and its cleartext default means anything that can reach the segment can read or inject traffic. Treat Industroyer2 as a concrete worked example, not a hypothetical, when prioritizing segmentation and monitoring: a real-world attacker used nothing more exotic than legitimately-formatted IEC-104 ASDU commands to open breakers, which is exactly what network segmentation and anomaly detection on unexpected command sequences are meant to catch. Re-check configuration after every firmware update, since a patch addressing one CVE is not a substitute for confirming TLS and authentication settings, and network segmentation, are still intact afterward.
What are the common pitfalls?
- Reading old vendor material as current. Datasheets that describe "challenge/reply" and "aggressive mode" describe the 2013 TS generation, not IEC 62351-5:2023.
- Treating TLS as a patch. CVE-2026-1773's advisory says 62351-3 "does not remediate the vulnerability but mitigates the risk of exploitation."
- Forgetting bidirectional mode. Two RTU500 CVEs apply only when IEC-104 bidirectional operation is configured, so configuration records matter as much as version numbers.
- One SBOM line for the whole stack. The 104 library, the TLS library and XML parsers used by co-located IEC 61850 functions are patched on different schedules; see IEC 61850 security.
- Licence surprises. lib60870 is dual-licensed GPLv3 or commercial; record which one each product ships under.
A passive capture from an IEC-104 segment's mirror port is a practical way to start the inventory in step 1 without connecting anything to a live protection circuit. SoftWSP's OT Snapshot builds a host inventory from that capture with the ports each host serves, so sessions on TCP 2404 and 19998 show who talks to whom and which connections arrive from outside the site; it does not decode IEC-104 ASDUs. Where make, model and firmware are confirmed, an engineer matches them against CISA ICS advisories and NVD, and the AI-drafted report cites each finding for engineer review. For device makers, a firmware security analysis extracts a CycloneDX 1.6 SBOM from a firmware image and matches components such as lib60870 or mbedtls against NVD, CISA KEV, FIRST EPSS and CSAF advisories.
This guide draws on IEC's own webstore catalogue and abstracts for IEC 60870-5-104, IEC 62351-3, IEC 62351-5 and IEC TS 60870-5-7, the publisher previews of three of them, the IANA port registry, a peer-reviewed IEEE PES paper, NIST SP 800-82 Revision 3, the CSAF files of seven CISA ICS advisories, the lib60870 repository and advisories, Siemens' hardening manual, and independent malware analyses from ESET, Mandiant and Netresec. The full paywalled IEC texts were not purchased for this guide; granular protocol-mechanism claims here are general description, and the primary IEC text is the authoritative source for a compliance or implementation decision. This guide is general information, not a substitute for a current CISA ICS advisory search for your exact hardware and firmware version.
Put this guide to work
Frequently asked questions
What is IEC 60870-5-104, and where is it used?
IEC 60870-5-104 — officially "Telecontrol equipment and systems – Part 5-104: Transmission protocols – Network access for IEC 60870-5-101 using standard transport profiles" — is an IEC Technical Committee 57 standard that extends the earlier serial protocol IEC 60870-5-101 over TCP/IP, registered by IANA on TCP/UDP port 2404 as "IEC 60870-5-104 process control over IP." It is widely described in industry practice as the European, international and Chinese counterpart to DNP3, which dominates North American grid telecontrol instead; both protocols serve the same role, moving monitoring and control data between a control center and substations, remote terminal units (RTUs) and protection relays across a geographically spread-out power or water system. Where DNP3 shows up constantly in CISA ICS advisories for North American utilities, IEC-104 shows up the same way for European and internationally deployed grid equipment — most visibly, in recent years, in Hitachi Energy's RTU500 product line.
Does IEC 60870-5-104 have any built-in authentication or encryption?
No. Academic security research is explicit on this point: a 2014 IEEE PES paper on stateful intrusion detection for the protocol states plainly that "the IEC 60870-5-104 protocol transmits messages in clear text without any authentication mechanism," and independent survey literature since then attributes the protocol's main vulnerabilities to "the lack of data integrity, availability and authentication mechanisms in the protocol's specification" — citing documented man-in-the-middle research as a direct consequence. NIST SP 800-82 Revision 3 does not name IEC 60870-5-104 specifically in the sections this guide could verify, but its general description of OT protocols applies directly: "Many OT protocols have no authentication at any level. Without authentication, there is the potential to replay, modify, or spoof data or devices." On the wire, IEC-104 running on its standard port 2404 is unencrypted TCP traffic that any device on the same network segment can read or, in the absence of transport security, inject into.
Is there a secure version of IEC 60870-5-104, and what does it actually protect?
Yes, as two separate add-on standards from the same IEC 62351 series that secures other TC 57 protocols. IEC 62351-3 ("Communication network and system security – Profiles including TCP/IP," current Edition 2.0 published 2023-06-06) defines a TLS 1.2/1.3 profile that wraps IEC-104's TCP/IP traffic in transport-layer encryption and integrity protection, typically run on a separate port (commonly cited as 19998) from the standard unencrypted port 2404. IEC 62351-5 ("Security for IEC 60870-5 and derivatives," promoted from a Technical Specification to a full International Standard, Edition 1.0, published 2023-01-13) adds application-layer security specific to IEC 60870-5 protocols: per IEC's own abstract, the 2023 edition removes the earlier Technical Specification's challenge/reply procedure, replaces its "aggressive mode" with a Secure Data message exchange run per controlling/controlled station association, and adds authenticated encryption of application data. Vendor documentation (Siemens' SICAM/SIPROTEC system hardening manual is one example) shows real products combining both — TLS for transport confidentiality plus 62351-5 / IEC 60870-5-7 application security — but CISA advisories against Hitachi Energy's RTU500 also show the security features carrying their own flaws: two 2022 CVEs could only be triggered when 62351-5 or 62351-3 support was configured. Whether a given RTU or control-center product has either standard enabled, and configured correctly, has to be verified against that specific product, not assumed from the standard's existence.
Has IEC-104 actually been used in a real grid attack?
Yes, documented directly by multiple independent security research teams analyzing the same malware. ESET's 2022 analysis of Industroyer2, the malware used against a Ukrainian energy provider, states plainly: "Industroyer2 only implements the IEC-104 (aka IEC 60870-5-104) protocol to communicate with industrial equipment. This includes protection relays, used in electrical substations." Mandiant's independent analysis of the same malware (tracked as INDUSTROYER.V2) confirms the mechanism in more technical detail: the malware "crafts configurable IEC-104 Application Service Data Unit (ASDU) messages... to change the state of a remote station's Information Object Addresses (IOAs) to ON or OFF" — in effect, sending legitimate-looking IEC-104 commands to open or close circuit breakers, something the protocol's lack of authentication does nothing to stop if an attacker can reach the network segment. Netresec's packet-level analysis of captured Industroyer2 traffic identifies the exact ASDU type IDs used (46 for double command, 45 for single command) against specific breaker information object addresses. The original 2016 Industroyer malware against the Ukrainian grid is independently reported (via ESET and Dragos research) to have included a similar IEC-104 payload module among several protocol-specific modules — cited here as well-corroborated secondary reporting rather than a directly re-checked primary quote.
Sources
- IEC: IEC 60870-5-104:2006 (Ed. 2.1 consolidated with AMD1:2016) — Telecontrol equipment and systems, Part 5-104
- IANA Service Name and Transport Protocol Port Number Registry — "iec-104", TCP/UDP port 2404
- Yang, Y. et al., "Stateful Intrusion Detection for IEC 60870-5-104 SCADA Security," IEEE PES General Meeting 2014
- NIST Special Publication 800-82 Revision 3, Guide to Operational Technology (OT) Security (September 2023)
- IEC: IEC 62351-3:2023, Part 3 — Communication network and system security – Profiles including TCP/IP
- IEC: IEC 62351-5:2023, Part 5 — Security for IEC 60870-5 and derivatives
- ESET: "Industroyer2: Industroyer reloaded" (12 April 2022)
- Mandiant / Google Cloud: "INDUSTROYER.V2: Old Malware New Tricks" (2022)
- CISA ICS Advisory ICSA-26-155-04: Hitachi Energy RTU500 (IEC 60870-5-104 bidirectional mode NULL pointer dereference)
- IEC: IEC TS 60870-5-7:2025 (Ed. 2.0) — Security extensions to IEC 60870-5-101 and IEC 60870-5-104 protocols (applying IEC 62351)
- IEC: IEC TS 60870-5-7:2013 (Ed. 1.0, revised) — Security extensions to IEC 60870-5-101 and IEC 60870-5-104 protocols
- IEC 62351-5:2023 publisher preview (foreword change list, contents, clause 5 problem description)
- IEC 60870-5-104:2006 publisher preview (contents: APCI, I/S/U formats, port number, k parameter)
- Netresec: "Industroyer2 IEC-104 Analysis" (April 2022)
- arXiv 2109.03945: "Vulnerabilities and Attacks Against Industrial Control Systems and Critical Infrastructures" (survey)
- Siemens: SICAM / SIPROTEC System Hardening for Substation Automation and Protection (security manual)
- MZ Automation lib60870 repository (IEC 60870-5-101/104 library, README and licensing)
- lib60870 v2.4.1 release notes (2026-07-15)
- lib60870 security advisory GHSA-7v97-jmwv-w5j7 (heap buffer over-read in InformationObject_ParseObjectAddress, patched in 2.4.1)
- cisagov CSAF: ICSA-26-211-11 MZ Automation lib60870 (CVE-2026-61893, CVE-2026-63033)
- cisagov CSAF: ICSA-26-155-04 Hitachi Energy RTU500 (CVE-2026-8479)
- cisagov CSAF: ICSA-26-062-03 Hitachi Energy RTU500 (CVE-2026-1773)
- cisagov CSAF: ICSA-25-093-01 Hitachi Energy RTU500 (CVE-2024-11499, CVE-2024-12169)
- cisagov CSAF: ICSA-25-128-02 Hitachi Energy RTU500 (CVE-2023-5769)
- cisagov CSAF: ICSA-24-354-01 Hitachi Energy RTU500 (CVE-2023-6711)
- cisagov CSAF: ICSA-23-220-02 Hitachi Energy RTU500 (CVE-2022-2502, CVE-2022-4608)
About this guide
SoftWSP OT/ICS Desk
Covers the OT/ICS security cluster: 87 guides.

In this cluster
Questions about this page?
