September 6, 2026 · 8 min read
TL;DR: The five modes are withdraw, smooth, compromise, force, and collaborate. They differ in who makes the decision. Collaborating gets the most durable outcome and is the default when nothing constrains you. Forcing is right when the scenario hands you positional power, an urgent call, or a safety and compliance line. A conflict that keeps recurring is the sign that previous rounds were accommodated rather than decided.
Ask most candidates for the best conflict resolution approach and they will say collaborating, which is right often enough to be dangerous. The exam builds whole question clusters around scenarios where collaborating is offered, is genuinely reasonable, and is still not the keyed answer, because the stem quietly told you something that rules it out.
The useful way to hold them is not by name but by who ends up making the call.
| Mode | Who decides | Durability | When the exam wants it |
|---|---|---|---|
| Withdraw / avoid | Nobody, yet | None; it returns | Cooling off, or the issue is genuinely someone else's |
| Smooth / accommodate | The other party | Low; underlying issue survives | The relationship matters more than this point |
| Compromise / reconcile | Both, partially | Medium; nobody fully satisfied | Deadlock with equal power and time pressure |
| Force / direct | You, on authority | Low commitment, high speed | Urgency, safety, compliance, or explicit positional power |
| Collaborate / problem-solve | Both, fully | Highest; both own it | The default when nothing above applies |
Compromise is the one candidates over-rate as a "nice" answer. It is lose-lose in the sense that matters: both parties give up something real, and neither is committed to the result. It has a place when time is short and the parties are evenly matched, but it is not a softer collaborating.
Common trap: reading win-win as splitting the difference. It is not. Win-win rests on believing an outcome exists that serves both parties and being willing to hunt for it, which takes enough character to hear a criticism without defending, enough trust that the other side says what they actually need, and the discipline to understand their position before arguing your own. When a functional manager complains that they were not properly engaged and the program manager asks you to handle it win-win, the keyed answer in this bank is to demonstrate integrity, build mutual trust, and look at it from the manager's point of view. Options that meet halfway, or that produce evidence you did engage them, both read as reasonable and both miss the mode being asked for.
More often than the "collaborate always" instinct expects, and the stem tells you.
The clearest case is when the question hands you the authority explicitly. A team drawn from many parts of the organisation, each with its own agenda, frequent small conflicts, very little work getting done, and a program manager who asks you to use your positional power. Forcing is the mode that runs on positional power. Its single advantage is speed, bought at the cost of commitment, because the side that loses complies rather than agrees. On a team fragmented by competing agendas and stalled, that trade is worth making, and the question is testing whether you can identify a mode from its mechanism rather than ranking the modes by how pleasant they sound.
Two other patterns where forcing wins: a safety or regulatory line, where there is nothing to negotiate about, and genuine time criticality, where the cost of a good decision arriving late exceeds the cost of a fast decision arriving unloved.
The mirror image is just as testable. On a project outside the project manager's technical field, two engineers argue heatedly over how to build a requirement. Facilitating so the team members themselves reach the decision is the answer, and the reason is competence rather than politeness. Deciding a technical question you are not qualified to decide is forcing on the strength of a job title, and the decision will be worse and worse implemented.
Because nothing was decided the last three times.
Here is the scenario the bank uses, and it repays close reading. A bus operator is fitting new ticketing hardware across its fleet. Depot engineering and the scheduling team have argued three times in four months about when vehicles come off service for the work. Each time, scheduling gave way to keep the peace. Each time, the argument came back.
The recurrence is the evidence. Every round ended with somebody accommodating, which stopped that round without settling anything, so it returns as soon as the next batch of vehicles is due. The answer is to set a standing rule for when vehicles are released and hold the programme to it. That decides it once and gives the programme something to plan against. Conceding again repeats what has already failed three times. Logging the disagreement and watching for a pattern records something you already know.
That pattern generalises: when a stem tells you a conflict has recurred, it is telling you the previous mode did not work, and you should be looking for the option that decides rather than the option that defuses.
One more distinction worth holding. Not every interpersonal problem is a conflict to resolve. When a stakeholder is known for losing their temper and the project manager wants to limit the effect on the team, the answer is emotional intelligence training. What the team is up against is emotional, not procedural. Negotiation and communication skills help reach agreements; they do nothing for a team's composure under pressure. Wrong answers here reliably offer the procedural fix for an emotional problem, and it is the same reasoning error that sends candidates to a change control board when what they need is a conversation.
What are the five conflict resolution modes on the PMP exam? Withdraw or avoid, smooth or accommodate, compromise or reconcile, force or direct, and collaborate or problem-solve. They differ in who makes the decision and how durable the outcome is.
Is collaborating always the right answer? No. Forcing is correct when the stem hands you positional power, or when safety, compliance, or genuine urgency removes the room to negotiate. Smoothing can be right when the relationship outweighs the point at issue. Read the stem for the constraint before reaching for collaborate.
What should a project manager do when a technical argument is outside their expertise? Facilitate and let the subject-matter experts decide. Imposing a technical call you are not qualified to make is forcing on a job title, and it produces a worse decision that is worse implemented.
Why does the same conflict keep coming back? Because each round was accommodated rather than decided. A recurring conflict calls for a standing rule or decision, not another round of giving way.
The 70 conflict questions in PMP Practice sit inside a bank re-certified against PMBOK 8 and the July 2026 Exam Content Outline, and every wrong answer carries the reasoning that makes it wrong, explained rather than just marked. Start the free 20-question sample with no card and no signup.
Related: Servant leadership vs traditional project management · Tuckman's stages of team development · People study guide
Sources: PMI — PMP Examination Content Outline, 2026 (PDF) · PMI — PMBOK Guide standards
What are the five conflict resolution modes on the PMP exam?
Withdraw or avoid, smooth or accommodate, compromise or reconcile, force or direct, and collaborate or problem-solve. They differ in who makes the decision and how durable the outcome is. Collaborating produces the most durable agreement because both parties own it; forcing produces the fastest one because a single authority imposes it.
Is collaborating always the right answer?
No, and questions that punish that reflex are common. Forcing is correct when the stem tells you to use positional power, when safety or compliance leaves no room to negotiate, or when a decision is genuinely urgent. Smoothing can be right when the relationship matters more than the point at issue. The stem usually contains the constraint that picks the mode; read it before reaching for collaborate.
What should a project manager do when a technical argument is outside their expertise?
Facilitate, and let the subject-matter experts on the team make the call. Imposing a technical decision you are not qualified to make is forcing on the strength of a role rather than knowledge, and the team will neither trust nor implement it well. The project manager owns that the decision gets made; the experts own what it is.
Why does the same conflict keep coming back?
Usually because each round ended with someone giving way rather than something being decided. Accommodating stops that instance of the argument without settling the underlying question, so it returns on the next occasion that raises it. A repeated conflict is a signal to set a standing rule or decision, not to smooth it over one more time.