Join our Newsletter — 33% off our NHI Course
Home› Guides› RDP Lateral Movement Forensics Guide: Source and Destination…
Guide Incident Response & Forensics

RDP Lateral Movement Forensics Guide: Source and Destination Artefacts, Bitmap Cache and Event IDs

← All guides
By NHI Mgmt Group, edited by Lalit Choda Updated 20 November 2026 14 min read
On this page

RDP lateral movement forensics reconstructs each Remote Desktop hop by pairing evidence from the source host, where the client ran, with evidence from the destination host, where the session landed. Remote Desktop Protocol is one of the most common ways attackers move between Windows systems once they hold valid credentials, because it is legitimate, usually allowed through internal firewalls and looks like normal administration. ATT&CK tracks it as T1021.001, Remote Desktop Protocol, under the Lateral Movement tactic.

This guide is for analysts who need to prove or disprove specific RDP hops during an investigation. It explains which artefacts sit on each side of a connection, what the Security log events Microsoft documents actually record, how the RDP client bitmap cache can show what an attacker saw on screen, and how to assemble the evidence into a hop record that stands up in a report. It is detection and investigation guidance only. For the incident where RDP was exposed to the internet, see the Exposed RDP Compromise guide; for hardening, see the Remote Desktop (RDP) Security Guide.

Event IDs, field names and paths on this page come from Microsoft documentation, CISA guidance and the documentation of the open source tools named. Behaviour varies between Windows versions and configurations, so confirm each artefact in your own lab against your own builds before relying on its absence.

Key takeaways

  • Every RDP hop has two sides. Destination evidence shows that a session arrived; source evidence shows who started it and from which user profile. Pair both wherever you can.
  • On the destination, event 4624 with Logon Type 10 (RemoteInteractive) is the core record, with Source Network Address and Workstation Name pointing at the source.
  • Events 4778 and 4779 record reconnects and disconnects with Session Name, Client Name and Client Address, which helps you see long running or repeatedly resumed sessions.
  • On the source, the per-user Terminal Server Client registry key holds the MRU list of computers connected to, and the client bitmap cache can hold image tiles of what was displayed.
  • Bitmap cache tiles are fragments, not screenshots. Treat reconstructed images as supporting evidence and describe them carefully.
  • Attackers use valid accounts, so the account in each hop needs scoping, including any service or shared administration accounts.

How RDP evidence is distributed

Think of an RDP hop as a conversation between two machines with a user profile on each end. The source host runs the Remote Desktop client in the context of a user profile, which may be the attacker's compromised account on that host. The destination host authenticates the incoming connection, creates or reconnects a session, and runs whatever the attacker does inside it.

That split matters for three reasons:

  • Different accounts. The account logged on at the source may differ from the account used to authenticate to the destination. Both are part of the hop.
  • Different retention. Destination Security logs on busy servers may roll over quickly, while per-user registry and cache files on the source can persist for a long time. Either side may be the only survivor.
  • Different questions. Destination evidence answers "who arrived here and from where". Source evidence answers "where did this user go from here", which is how you find the next hop.

In a typical intrusion you work outwards from a known compromised host: destination evidence tells you where the attacker came from, source evidence tells you where they went next. The Lateral Movement Guide covers the wider technique set, and the Lateral Movement Artefacts Cheat Sheet gives a fill-in table to record which artefacts you have confirmed per method.

Destination host: Security log events

The Windows Security log on the destination is usually the first stop. Microsoft documents the following events and fields.

4624: An account was successfully logged on

Microsoft's logon type table describes Logon Type 10, RemoteInteractive, as "A user logged on to this computer remotely using Terminal Services or Remote Desktop". Logon Type 12, CachedRemoteInteractive, is described as the same as RemoteInteractive and used for internal auditing. Logon Type 7, Unlock, can also appear around RDP activity when a session is resumed, so record when your builds produce it.

Fields that matter for RDP:

  • Source Network Address: the IP address of the machine from which the logon attempt was performed. Microsoft notes that ::1 or 127.0.0.1 means localhost, which can indicate tunnelling or local forwarding and should prompt further review.
  • Workstation Name: the machine name from which the logon attempt was performed. Compare it with the source address; a mismatch with your asset inventory is worth noting.
  • Logon ID: lets you correlate this logon with later events in the same session, such as process creation.
  • Linked Logon ID: the paired logon session, or 0x0 if none.
  • Restricted Admin Mode: populated only for RemoteInteractive logons, a Yes or No flag showing whether credentials were passed using Restricted Admin mode.
  • Elevated Token: Yes if the session has administrator privileges.
  • Authentication Package: commonly NTLM, Kerberos or Negotiate.

4778 and 4779: session reconnected and disconnected

Event 4778, "A session was reconnected to a Window Station", is generated when a user reconnects to an existing Terminal Services session or switches to an existing desktop using Fast User Switching. Event 4779 records a session disconnected from a Window Station. Both carry:

  • Session Name: Microsoft gives RDP-Tcp#N as the typical RDP session name and Console as typical for Fast User Switching.
  • Client Name: the computer name the user connected or disconnected from, or "Unknown" for console sessions.
  • Client Address: the IP address of that computer, or "LOCAL" for console sessions.
  • Logon ID: correlates with the original 4624 event.

These events are particularly useful when an attacker leaves a session disconnected and resumes it later, perhaps from a different source. A single 4624 followed by several 4778 events from different client addresses tells a different story from repeated fresh logons. Microsoft's monitoring guidance suggests watching for Session Name values containing RDP-Tcp# where RDP is not allowed for specific users or computers, and monitoring Client Address to confirm connections only come from expected internal addresses.

Terminal Services operational logs

CISA's #StopRansomware Guide advises reviewing the TerminalServices-RemoteConnectionManager event log to check for successful RDP network connections. Collect the Terminal Services operational logs alongside the Security log on every destination; they record connection activity from the Remote Desktop service's point of view and can survive when the Security log has rolled over. Confirm the event IDs and fields your Windows versions produce in a lab before building detections on them.

Activity inside the session

Once you have the Logon ID of an RDP session, use it to find what happened inside. Event 4688, "A new process has been created", records the New Process Name and, on newer versions, the Creator Process Name. The Process Command Line field is empty unless "Include command line in process creation events" is enabled, so confirm audit policy first. Other destination artefacts such as program execution and file access records are covered in the Windows Forensic Artefacts Guide.

Source host: client registry evidence

On the source host, the Remote Desktop client stores per-user history. Microsoft documents that the Remote Desktop Connection computer box history is held under:

HKEY_CURRENT_USER\Software\Microsoft\Terminal Server Client\Default

Entries appear as MRU followed by a number. Microsoft states that every new connection is given the value MRU0 and the other values move down in number, and that an MRU value can contain a fully qualified domain name or an IP address of the computer connected to. Microsoft's examples are MRU0 set to an IP address and MRU1 set to a host name.

Investigative use:

  • Next hop discovery. The MRU list on a compromised source tells you which destinations that user profile connected to, which gives you a list of hosts to examine next.
  • Ordering. MRU0 is the most recent. The order shows relative recency, not exact times.
  • Timing. The registry key's last written time records when the key last changed, which may relate to the most recent entry. Treat it as an indicator only and corroborate with destination logs.
  • Profile context. Because the key sits under HKEY_CURRENT_USER, it belongs to a specific user profile. Collect the registry hive for each profile on the source host, not only the one you expect.

Absence is weak evidence. Users can clear the history, and connections launched in other ways may not populate it. Microsoft's article is itself a guide to deleting these entries, which tells you a careful attacker could do the same.

Also look for process execution of the Remote Desktop client on the source, with command lines if logged, and for files that launch connections. Process creation records from the source, tied to the same user, help put times on the hop.

Source host: the RDP bitmap cache

The Remote Desktop client can keep a cache of bitmap data from the remote session on the client machine. For investigators, this means the source host can hold image fragments, or tiles, of what an attacker saw on the destination's screen.

The ForensicArtifacts project, a community catalogue of forensic artefact definitions, defines WindowsRDPClientBitmapCache as "Artifacts of RDP connection contents" with the path:

%%users.localappdata%%\Microsoft\Terminal Server Client\Cache\*.*

That is the Terminal Server Client cache folder under each user's local application data. The ANSSI bmc-tools project, an open source RDP bitmap cache parser, states that it processes bcache*.bmc and cache????.bin files found inside Windows user profiles. Its documented options include extracting a given number of bitmaps, extracting old bitmap data found in BMCache files, producing a collage bitmap that aggregates all tiles and setting the number of tiles per line in that collage.

How to use bitmap cache evidence responsibly:

  • Collect per profile. Copy the cache folder from every user profile on the source host as part of triage, with hashes, before any reimaging.
  • Parse on an analysis system. Run the parser against copies on an isolated analysis workstation, never on the evidence host.
  • Review tiles and collages. Individual tiles are small. A collage can make fragments of windows, text, icons or command prompts recognisable, but tiles are not in display order and the image is not a screenshot.
  • Corroborate. If a tile appears to show a tool, a file name or a console window on a particular server, confirm it with destination logs and file system evidence before reporting it as fact.
  • Describe limits. The cache does not record which session or which destination a tile came from on its own, and it does not give precise times. Use file system timestamps of the cache files and the registry MRU to suggest context, and say so.
  • Handle sensitively. Tiles can contain personal or confidential data visible on the remote screen. Apply the same handling rules as for any other evidence containing sensitive data.

The bitmap cache is often the only place you can see what the attacker looked at, which can be decisive when scoping data access. It is also easy to forget, because it sits in a user profile rather than in a log. Add it to your standard collection profile for any host suspected as an RDP source.

Network and identity evidence

Host artefacts are strongest when combined with evidence from outside the hosts:

  • Network sessions. Firewall and flow logs showing connections to the RDP service port between the two hosts at the right times confirm the hop independently and help when host logs are gone. The Network Forensics Guide covers this.
  • Remote access gateways. If RDP is brokered through a gateway or bastion, its logs show the real user and client behind internal connections.
  • Domain controller authentication. Authentication events on domain controllers can show the account and source for a hop even when the destination's own log has rolled over.
  • Identity context. For each account seen, record whether it is a named user, a shared administration account or a service account, and whether multi-factor authentication applies. Shared and non-human accounts used for RDP are hard to attribute and should be flagged. The MFA for Windows Logon and RDP Guide covers prevention.

Building the hop record

For each suspected RDP hop, record one row in a hop table:

FieldWhat to record
Hop identifierA sequential reference used throughout the report
Source host and profileHost name, IP address at the time, and the user profile running the client
Destination hostHost name and IP address
Account usedThe account in the destination 4624 event and its type
Times (UTC)First logon, reconnects, disconnects and last activity
Destination evidence4624 Logon Type 10 records, 4778 and 4779 records, Terminal Services operational log entries
Source evidenceTerminal Server Client MRU entry, client process execution, bitmap cache findings
Independent evidenceNetwork flows, gateway logs, domain controller authentication
Activity in sessionProcesses and file access tied to the Logon ID
ConfidenceConfirmed, likely or possible, with the reason

Two sided evidence plus independent network evidence is a confirmed hop. One side alone is usually "likely". The table then feeds the master timeline described in the Forensic Timeline Analysis Guide.

Illustrative hunting logic

The following ideas are illustrative and need tuning to your environment. They use only the events and fields described above.

  • Event 4624 with Logon Type 10 where the Source Network Address is a workstation subnet and the destination is a server tier where administrators normally use a jump host.
  • Event 4624 with Logon Type 10 where Source Network Address is ::1 or 127.0.0.1, which can indicate local forwarding or tunnelling.
  • Event 4778 for one Logon ID with Client Address values that change across reconnects.
  • Logon Type 10 sessions by service accounts or other non-human identities, which should rarely log on interactively.
  • Terminal Server Client MRU entries on user workstations naming servers the user has no business reason to reach.

Common failure patterns

  • Only collecting destination logs. Without source evidence you cannot find the next hop or see what the attacker viewed.
  • Forgetting the bitmap cache. It sits in user profiles, not logs, and is lost when the source host is reimaged.
  • Reading tiles as screenshots. Overclaiming from fragmented images undermines the report.
  • Ignoring reconnects. Looking only at 4624 misses sessions that were resumed from new sources via 4778.
  • Trusting Workstation Name alone. Names can be misleading; pair them with the source address and inventory.
  • Collecting only one user profile. Attackers often use a different profile from the one you first suspect.
  • Not testing in a lab. Assuming an artefact appears on your builds without checking leads to false negatives.

Practitioner checklist

  • Collect Security and Terminal Services operational logs from every suspected destination.
  • Filter 4624 for Logon Types 10 and 12 and capture Source Network Address, Workstation Name, Logon ID and Restricted Admin Mode.
  • Collect 4778 and 4779 events and correlate them by Logon ID.
  • Export the registry hive and the Terminal Server Client cache folder for every user profile on each suspected source.
  • Parse bitmap caches on an isolated analysis system and corroborate any finding.
  • Add network flow, gateway and domain controller authentication evidence for each hop.
  • Record every hop in a hop table with a confidence level.
  • Scope each account used, including shared and service accounts, and feed them into credential rotation.
  • Validate artefact behaviour on your own Windows builds in a lab.

Standards and references

Related NHI Mgmt Group resources: Exposed RDP Compromise · Remote Desktop (RDP) Security Guide · Lateral Movement Guide · Windows Forensic Artefacts Guide

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.

NHIMG Editorial Note
Researched and written by the NHI Mgmt Group editorial team using AI tools, under the editorial direction of Lalit Choda. Facts are drawn from the sources listed on this page and checked by our automated quality process. Last updated 20 November 2026. Spotted an error? Use our contact page.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM and Agentic AI security. nhimg.org