TOEFL Adaptive Test 2026: How Stage 1 & Stage 2 Routing Sets Your Score Ceiling

If you opened the new TOEFL iBT in 2026 and felt the test was reading you, that is because it was. Reading and Listening are now multistage adaptive, which means the second half of each section is chosen based on how you handled the first half. That single design change quietly decides what your maximum possible band will be on the day, and most students do not know it until after they have already capped themselves out. This guide unpacks exactly how the routing works, what the hidden score ceiling is, and the practical strategy for keeping band 6 within reach.

1. What "adaptive" actually means on the 2026 TOEFL

Multistage adaptive testing, or MST, is a delivery format where the exam decides which set of questions to give you next based on how you performed on the previous set. The 2026 TOEFL uses a two-stage version of MST in Reading and Listening only. You first complete a fixed-difficulty routing module, then the test selects one of two follow-up modules for you: a harder one or an easier one.

This is different from item-by-item adaptive tests like the GRE, where every single question is chosen for you in real time. On the new TOEFL, adaptation happens at the module level, not at the question level. The benefit for ETS is that it produces more precise scores at the high and low ends of ability without needing a 60-question section.

Speaking and Writing remain entirely linear. Every test taker sees the same Listen & Repeat items, the same interview structure, and the same writing tasks. We will come back to why ETS made that choice.

2. Stage 1 vs Stage 2: how routing works

Each adaptive section opens with the routing module. In Reading, that is one passage with roughly 10 questions. In Listening, it is a fixed bundle of items spanning a conversation and a lecture. You have no idea you are being routed; the interface looks identical to a normal section.

When you submit the routing module, the scoring engine evaluates your answers and assigns you a path. Two paths exist:

ETS does not publish the precise routing threshold. Based on documented MST behavior across other standardized exams that use a similar two-stage design, the threshold tends to sit around the 60 to 70 percent mark on the routing module, weighted by item difficulty. Getting an easy item wrong hurts your routing more than getting a hard item wrong, because the model expects you to handle the easy ones.

3. The score cap nobody tells you about

Here is the part you do not see in the official ETS materials. Once you are routed to the easy Stage 2 path, your score is mathematically capped well below the top of the scale. A perfect performance on the easy second module will not lift you above approximately band 4.0 on that section.

This is not punitive. It is a built-in property of MST. The easy module simply does not contain enough hard items to demonstrate band 5 or band 6 ability. If you only ever answer questions calibrated at the B1 to B2 level correctly, the model has no evidence you can handle C1 or C2 material, and so it will not certify you at that level.

The practical consequence is brutal: a strong student who has a bad opening 10 questions in Reading can cap themselves at band 4 on that section before they have even seen Stage 2. No matter how flawlessly they handle the rest of the test, the ceiling is set. We see this most often in test takers who got nervous and rushed the first passage, or who second-guessed themselves on the routing module's easier items.

4. Why Speaking and Writing are not adaptive

Adaptive routing only works when items can be auto-scored against a calibrated difficulty model. Multiple choice and select-the-correct-option items fit that model perfectly. Open-ended responses do not. Speaking and Writing tasks are graded by human raters and AI-assisted scoring against a six-point band rubric, which means there is no programmatic way to decide mid-section whether to give you a harder prompt next.

This has two implications for your prep. First, Speaking and Writing have no hidden score ceilings. Whatever band you earn is what your performance demonstrated, full stop. Second, you cannot rely on a strong Reading or Listening band to lift a weak Speaking or Writing band. Each section is reported separately and the overall band score is the average of all four. There is no compensatory scoring on the new TOEFL.

5. How adaptive scoring maps to the 1-6 band scale

The 2026 TOEFL reports each section on a 1.0 to 6.0 band scale, in 0.5 increments, with each band aligned to a CEFR level. The adaptive engine combines two pieces of information to land on your final number: which Stage 2 module you were routed to, and how many items you got right within that module.

The two ranges overlap deliberately around band 3.5 to 4.0. That overlap is a quality check: students who land in that band can be measured accurately by either module, so the routing decision matters less in the middle of the scale than at the extremes. If you are aiming for band 4.5 or higher, the routing is decisive. If you are comfortable with band 3.5, it is forgiving.

6. Stage 1 strategy: the highest-leverage 10 questions

Treat the routing module like the most important block of questions on the entire test, because it is. Three habits matter most.

• 1 Bank time, do not spend it. Aim to finish Stage 1 in slightly less than half your section time. The hard path of Stage 2 contains denser passages and longer audio, and you will need every minute. Walking out of Stage 1 with two minutes saved is a win.
• 2 Do not rush past the easy items. Routing weights penalize easy-item errors more than hard-item errors. The model assumes any test taker capable of the hard path should sweep the easy items. One careless mistake on a vocabulary question can be the difference between paths.
• 3 Use the review feature inside Stage 1. You can return to and change any answer within the routing module before submission. Use the last 90 seconds to revisit anything you flagged. Once Stage 2 begins, those answers are locked.

7. Common mistakes that send strong students down the easy path

These are the failure patterns we see most often when reviewing students' practice attempts. Each one is fixable.

8. How to practice for adaptive routing

Most free TOEFL practice materials online are still linear. A linear practice test cannot reproduce the felt pressure of routing or the consequence of mistakes in the opening module, so students often arrive at the real test having never trained for the moment that matters most. There are three things you can do about that.

First, take adaptive-format practice tests when you can. Our free TOEFLMock practice tests simulate the two-stage routing structure for both Reading and Listening, including the path-dependent score cap, so you can experience the routing decision before exam day.

Second, time-box your first 10 questions in any practice session. Even on linear material, force yourself to treat the opening 10 items as if they were a routing module. Build the habit of pacing-and-bank.

Third, review your routing-module errors in extreme detail. After every practice attempt, reread every Stage 1 item you missed and write down whether it was a comprehension issue, a careless reading, or a time-pressure error. Patterns will emerge within three or four sessions, and they tell you exactly where to invest your prep time. For deeper section strategy, see our guides on Reading section tips and Listening section tips.

9. FAQ

The adaptive redesign is the single most consequential change in the 2026 TOEFL, and it rewards the students who understand it. Treat the routing module as your most important block of questions, bank time, do not chase trick interpretations of easy items, and use practice that actually simulates routing. Do that and the score cap stops being something that happens to you and becomes something you decide.