Comparing Pass-the-Hash and Overpass-the-Hash

The same captured hash can lead to different authentication questions. Comparing pass-the-hash and overpass-the-hash means deciding whether the test uses NTLM material directly or uses it to establish Kerberos context. In this lab, a second machine is a possible destination, but…

The same captured hash can lead to different authentication questions. Comparing pass-the-hash and overpass-the-hash means deciding whether the test uses NTLM material directly or uses it to establish Kerberos context. In this lab, a second machine is a possible destination, but neither authentication nor access has been proved. We will distinguish the two mechanisms, follow their intermediate results, and verify the identity and action accepted at the destination.

Lateral movement means using access or credentials associated with one part of an environment to reach another system or security context. Pass-the-hash and overpass-the-hash are two credential-based approaches that can contribute to that movement. This lesson stays with the distinction between the two hash-based approaches. The separate details of those other remote-execution mechanisms belong to their own analysis.

Before doing anything, define the question. Perhaps it’s, “Can this captured hash authenticate the associated account directly to the second machine through an NTLM-based path?” That question points toward pass-the-hash.

A different question would be, “Can this hash material establish Kerberos authentication context for the account?” That points toward overpass-the-hash. These aren’t interchangeable phrasings. They name different authentication paths, which means they have different dependencies and can produce different evidence.

Your authorization boundary matters just as much as the technical distinction. Confirm that both systems are in the deliberately vulnerable lab or another environment you’re explicitly permitted to test. Identify the source of your current access, the target you’re considering, the account associated with the hash, and the exact action you’re allowed to perform. Credential reuse can affect systems beyond the machine where the material was obtained. An unclear boundary is a reason to pause, not an invitation to discover where the credential works.

Now establish a baseline. Record whether the target is reachable through the relevant network path and whether the intended remote service appears available. Record your current identity and access level.

Then write down what you expect to observe if the test succeeds. A useful prediction might say that the target should accept the account through the selected authentication path and that a separate check should show the resulting identity or access context on that target. The prediction should also say what would count as rejection or ambiguity.

This preparation prevents a common mistake. Knowing that a tool accepts a hash as input doesn’t establish that pass-the-hash is appropriate, and seeing a positive-looking message doesn’t establish lateral movement. A tool can process input successfully while the remote authentication fails.

Authentication might succeed without the account having permission to perform the requested action. A remote action might begin without giving you the identity or level of access you expected. Each transition needs evidence of its own.

Pass-the-hash is the direct reuse of captured password-hash material for an authentication flow that can use it, commonly discussed in relation to NTLM authentication. NTLM is a Windows authentication technology. In this technique, you don’t recover or type the account’s plaintext password. The hash material is used directly in the authentication process. The practical question is whether the target accepts that authentication for the claimed identity and whether that identity is authorized for the requested remote access.

That last distinction is essential. Authentication answers whether the target accepts the identity claim through the attempted mechanism. Authorization answers what the accepted identity may do. A valid hash isn’t a universal remote-administration token.

The relevant service must be reachable and willing to use the authentication path. The account must be recognized in the applicable identity context. It must also have enough permission for the requested operation. Network controls, authentication policy, service configuration, and account privileges can independently interrupt the chain.

Suppose the lab contains workstation Alpha, server Bravo, and a domain account called Lab Operator. You’ve obtained the account’s hash material from Alpha under the lab rules. Your baseline shows that Bravo is reachable and that an intended remote service is exposed.

A bounded pass-the-hash test asks Bravo to authenticate Lab Operator through the direct hash-reuse path. If Bravo rejects authentication, you haven’t proved that the hash is wrong. You’ve observed only that this specific attempt, against this target and service, didn’t establish the expected authentication.

If authentication appears to succeed, you still need to validate the result. Confirm the target identity so you don’t mistake a local effect on Alpha for access to Bravo. Confirm the resulting account or security context.

Then perform only the minimum authorized action needed to demonstrate the granted access. A tool’s success banner may support your interpretation, but it shouldn’t carry the entire conclusion. Target-side identity and context evidence are stronger than an isolated message from the initiating tool.

Overpass-the-hash uses hash material differently. Instead of treating direct NTLM hash reuse as the final authentication route to the target service, it uses the material to obtain or establish Kerberos authentication context for the account. Kerberos is a ticket-based authentication system widely associated with Active Directory environments. The immediate objective is therefore not simply, “Did the remote service accept this hash through NTLM?” It’s, “Did the supplied material result in usable Kerberos context for the intended identity?”

That difference changes what you need to validate. For pass-the-hash, the significant evidence follows the direct hash-based authentication attempt to the target. For overpass-the-hash, an intermediate change matters.

You need evidence that the expected Kerberos context was established, followed by evidence that a later request actually used that context as intended. A remote-access result by itself may not explain which authentication path produced it. If you don’t validate the context, you may label ordinary access, cached state, or another available credential as overpass-the-hash.

Return to Alpha, Bravo, and Lab Operator. Imagine that direct NTLM authentication is rejected by an environmental control, but the lab’s Kerberos path remains available and the account material can establish the required context. An overpass-the-hash test could then have a different result from pass-the-hash.

Reverse the conditions and the opposite may occur. If the Kerberos prerequisites aren’t met, overpass-the-hash can fail even when direct NTLM hash reuse is possible. Success or failure on one path doesn’t prove the state of the other.

Pass-the-hash tests an NTLM authentication path using suitable hash material. Overpass-the-hash introduces a Kerberos-context step before later service access. A tool accepting the input proves neither authentication nor the destination operation.

This is why environmental controls need names and observations. Saying “security blocked it” is too vague. Was the target service unreachable?

Was the selected authentication method unavailable or restricted? Did the target reject the identity? Was Kerberos context not established?

Did authentication succeed but authorization prevent the requested remote action? Each explanation describes a different break in the chain. Your evidence may not always reveal the exact cause, but it should let you distinguish what you observed from what you suspect.

Read output in layers. First, note what the initiating tool directly reports. That might be a connection failure, authentication rejection, access denial, or apparent success. Next, identify what the target or resulting session lets you verify.

Then state the narrowest conclusion supported by both. “The client reported successful authentication” is an observation. “The account obtained administrative execution on Bravo” is a stronger conclusion that requires identity, target, and privilege evidence. Don’t let the second statement ride on the first without validation.

A positive-looking indicator can still be incomplete. Perhaps the target accepted authentication, but the account lacked rights to the requested operation. That demonstrates something useful about the credential path, but it doesn’t prove successful remote execution.

Perhaps a process started remotely, but you haven’t established which identity owns it. That supports execution, but not the claimed account context. Perhaps a Kerberos-related artifact appears, but the subsequent access uses some other available context. That doesn’t yet prove the complete overpass-the-hash chain.

A failed result is equally easy to overstate. “Pass-the-hash is impossible here” is usually broader than one attempt can support. The attempt could have used the wrong target identifier, the wrong account context, an unsuitable service, a blocked network path, incorrect input, insufficient privilege, or a protocol state that didn’t match your assumption. Failure is evidence about the test you actually ran. It becomes evidence about a broader hypothesis only after you’ve isolated the relevant prerequisites.

Use a controlled workflow rather than changing several things at once. Begin with the baseline you already recorded. Choose either the direct NTLM reuse hypothesis or the Kerberos-context hypothesis.

Perform one bounded test, preserve the exact input and output, and compare the observation with your prediction. If the result is ambiguous, select the lowest-impact check that can isolate one uncertain prerequisite. Then adjust only the condition connected to that uncertainty before repeating the test.

Suppose a pass-the-hash attempt against Bravo returns an access-denied result. You first separate network reachability from authentication because reaching the service doesn’t mean the account was accepted. Then separate authentication from authorization because an accepted identity may still lack the requested administrative right.

Re-read the exact output instead of translating every denial into “bad hash.” If available evidence shows that authentication succeeded but the requested action was denied, changing hash formats or launching unrelated tools won’t address the demonstrated problem. The next decision concerns permission or a different authorized path.

Keep the source session separate from the destination result. New ticket material on the source is not yet an operation on the target, and a success banner may describe only one intermediate step. Verify the target and account associated with the access actually obtained.

Now suppose an overpass-the-hash attempt doesn’t produce the expected Kerberos context. Don’t immediately conclude that the account material is unusable everywhere. Recheck the identity you intended to represent, the target environment, the relevant authentication state, and whether the expected context can be observed.

If one prerequisite was assumed rather than measured, validate that prerequisite separately. Only then repeat the bounded attempt. This keeps troubleshooting attached to the mechanism instead of turning it into random command substitution.

Changing one assumption at a time also protects the value of negative evidence. If you alter the target name, account, protocol choice, tool, and requested action together, a later success won’t tell you which correction mattered. If you change only the target identifier and the expected Kerberos context then appears, you have a much clearer explanation. Preserve both attempts. The failed path documents the original condition, and the corrected path documents the evidence that justified your revision.

Manual understanding still matters when an authorized training tool can automate several stages. You should be able to describe whether the tool is attempting direct hash reuse or establishing Kerberos context, which system receives the authentication request, what identity is claimed, and what resulting evidence confirms access. Automation can accelerate a lab exercise, but it can also hide the boundary between authentication, authorization, and execution. If you can’t explain those boundaries, reproducing or troubleshooting the result becomes difficult.

A reliable record begins before the successful attempt. Note the date and relevant lab context, the source system, the intended target, and the authorization boundary. Identify the account without exposing credential material unnecessarily in the final report. Record how the hash material was obtained within the exercise, the authentication path being tested, and the prerequisite observations. Include the exact command or request you used in your working notes, along with meaningful output and any screenshots required for the assessment.

Then capture the result in terms another technically competent person can verify. State whether the observed path was direct hash reuse or a transition into Kerberos context. Name the target you actually reached.

Record the resulting identity and access level, the validation action, and the evidence that ties that action to the target. If the attempt failed, preserve the error, the assumption you tested next, and the reason for changing it. A report shouldn’t require you to remember what a vague screenshot was meant to prove.

Keep facts and interpretations visibly separate in your language. “Bravo accepted the authentication request and the resulting context identified Lab Operator” reports observations. “The environment may permit reuse of this account material on additional systems” is a hypothesis that would require separate authorization and testing. Likewise, failure against Bravo doesn’t prove that every target blocks the technique. Precise wording prevents a narrow observation from becoming an unsupported environment-wide claim.

This doesn’t make tools unimportant. It changes what competence looks like. You need enough understanding to recognize the authentication path, select a permitted implementation, interpret the evidence, and explain the result without treating the tool’s label as proof. If a training utility offers both pass-the-hash and overpass-the-hash features, the shared interface doesn’t erase the protocol distinction. Your notes should still identify which path you tested and how you verified it.

Once the result is validated, choose the next move from what the evidence actually established. Confirmed direct hash reuse may justify evaluating an authorized NTLM-capable access path where the demonstrated identity has relevant rights. Confirmed Kerberos context may justify an authorized request that can use that context. Successful authentication without sufficient authorization points toward permissions analysis, not repetition of the same request. An unreachable or blocked service points toward another permitted lateral-movement mechanism rather than a claim that the credential itself failed.

That next mechanism might involve a different Windows remote-management path, but it must be assessed on its own conditions. Don’t assume that success with pass-the-hash guarantees success through Windows Remote Management, Windows Management Instrumentation, PsExec, or DCOM. Those mechanisms can depend on their own services, permissions, network exposure, and configuration. Likewise, failure with one remote method doesn’t erase valid evidence about an authentication context that you independently confirmed.

The original decision is now more precise. You attempt pass-the-hash when you have authorization, relevant hash material, a plausible direct authentication path, and a clear way to validate the target and resulting identity. You consider overpass-the-hash when the question concerns establishing Kerberos context from that material and you can verify both the context and its later use. If those prerequisites or validation steps are missing, the right next action is additional observation, not a more aggressive tool.

Pass-the-hash uses NTLM material directly, while overpass-the-hash uses it to obtain Kerberos context; the resulting identity and destination access must be verified. Carry one question into your next practice session: Are you using NTLM material directly or converting it into Kerberos context, and what destination access resulted? The companion study reference includes a review checklist and an answer guide. This lesson was produced by Bare Metal Cyber dot com.

Comparing Pass-the-Hash and Overpass-the-Hash
Broadcast by