What Is Multistage Adaptive Testing?
Multistage adaptive testing (MST) is a test design in which test-takers move through several blocks (or “modules”) of items, and the difficulty of the later module is selected based on performance in an earlier module. The ETS 2026 TOEFL iBT blueprint confirms that Reading and Listening each follow this design.
In TOEFL 2026, every test-taker first answers a router module, then is placed into either a lower module or an upper module. Unlike item-level adaptive tests (where each individual item changes), MST operates at the module level — so the difficulty “step” happens once, between the router and the chosen module.
Why MST is used
A staged adaptive design lets the test target items more closely to a test-taker’s estimated ability after the router. This generally improves measurement efficiency, particularly at the upper and lower ends of the scale. ETS describes the design at a structural level in the blueprint; the precise psychometric model used to convert responses into the 1–6 band score has not been publicly released.
Writing and Speaking are linear, not adaptive. The blueprint states that all test-takers of a specific form receive the same set of Writing and Speaking tasks. Adaptive logic applies only to Reading and Listening.
What the ETS Blueprint Confirms (and What It Doesn't)
The April 2026 blueprint is one of the clearest public documents on TOEFL 2026 structure. It is useful to separate what ETS has formally confirmed from what remains undisclosed.
Confirmed by the blueprint
- Reading and Listening are two-stage adaptive: router → lower or upper module.
- Writing and Speaking are linear.
- Reading totals 50 items (router 18–21 min + module ~9 min); Listening totals 47 items (router 18 min + module 7–11 min).
- Some items in Reading and Listening are non-scored, used to calibrate future items.
- AI scoring applies to three tasks: Write an Email, Write for an Academic Discussion, and Take an Interview. The blueprint describes the AI model as evaluating features such as fluency, coherence, grammar, intelligibility, and communication effectiveness. Listen and Repeat, Build a Sentence, and all Reading and Listening items are machine-scored using rule-based logic against predefined answers.
Not disclosed by ETS
- The exact routing thresholds between the router and the lower/upper module.
- The exact psychometric algorithm used to compute the final 1–6 band score.
- How responses to non-scored items relate (if at all) to routing decisions.
- Whether — and how — different module paths interact with the highest achievable bands.
Router module — same for everyone
All test-takers start with the same router module. The ETS blueprint describes this as the shared entry stage for the Reading and Listening sections, lasting roughly 18–21 min (Reading) or 18 min (Listening).
Routing decision
After the router, the system places each test-taker into either the lower or upper module. ETS has confirmed that this routing depends on performance in the router, but the exact thresholds have not been publicly disclosed.
Lower or upper module
The upper module typically presents more challenging items; the lower module presents more accessible ones. Module length is approximately 9 min (Reading) or 7–11 min (Listening, depending on path).
Final band calculation
The final 1–6 band reflects performance across both the router and the assigned module. ETS describes this as a two-stage adaptive design but has not released the precise psychometric scoring algorithm. The blueprint also notes that some items are non-scored (used to calibrate future test material), which adds another factor that is not detailed publicly.
A careful note on score ceilings
Some MST-style assessments do constrain the maximum reportable score by module placement, but ETS has not confirmed whether TOEFL 2026 works this way. The blueprint confirms the structure but not the exact scoring algorithm. Until ETS publishes more detail, treat module placement as an indicator of router performance rather than a definitive ceiling on what is achievable. The most defensible takeaway: consistent accuracy in the router gives you the best chance of being placed into the module that matches your true ability.
Reading Under MST
Reading is delivered as a router module followed by either a lower or upper module. The blueprint specifies 50 total items (including some unscored items), with the router running approximately 18–21 minutes and the assigned module running about 9 minutes.
Reading Task Types
Complete the Words
30 items targeting B1–C1+. Construct meaningful text using word-ordering rules, lexical knowledge, and reading comprehension. Machine-scored.
Read in Daily Life
5–15 items targeting A1–C1. Short non-academic texts (signs, menus, emails, social posts). Machine-scored.
Read an Academic Passage
5–15 items targeting B1–C2. Longer academic texts requiring main idea, inference, and rhetorical structure recognition. Machine-scored.
What MST Means for Reading Preparation
The blueprint indicates that the upper module typically includes more complex vocabulary, denser inference questions, and items about rhetorical structure. To prepare for either path, build comprehension that goes beyond surface-level reading — practice identifying implicit meaning, author stance, and organisational structure.
Complete the Words items rely on contextual vocabulary and word-formation knowledge. Wide reading in both everyday and academic registers builds the lexical range that supports accuracy in either module.
See the TOEFL Reading guide for full task-type practice.
Listening Under MST
Listening uses the same router + module structure. The blueprint specifies 47 total items (including some unscored items), with the router running ~18 minutes and the assigned module running 7 minutes (lower) or 11 minutes (upper).
Listening Task Types
Listen and Choose a Response
15–19 items targeting A1–B2. Single-exchange dialogue between two people. Tests pragmatic, everyday listening.
Listen to a Conversation
10 items targeting A2–C1. Short conversations between two people in social/interpersonal contexts.
Listen to an Announcement
6–10 items targeting A2–C1. Classroom or campus-related announcements (academic navigational language).
Listen to an Academic Talk
8–16 items targeting A2–C2. Academic talks; tests main ideas, supporting details, inference, and sometimes idiomatic vocabulary.
What MST Means for Listening Preparation
Inference, speaker-attitude, and purpose questions tend to appear more frequently in the upper module of the listening blueprint. Active note-taking — capturing main point, supporting details, speaker stance, and signal words — is well aligned with both modules.
Listen and Choose a Response items reward intuition for natural-spoken English: register, tone, and the appropriate next reply in conversation. Exposure to authentic dialogue (podcasts, lectures, conversations) builds this intuition more reliably than memorising answer patterns.
See the TOEFL Listening guide for full task-type practice.
Router-Stage Strategy
The router module is the only stage every test-taker shares, and the blueprint identifies it as the input to the routing decision. ETS has not disclosed the exact thresholds, but as a structural matter, consistent accuracy in the router is the most reliable lever a test-taker has.
Key Router Principles
Don't rush the router for the sake of speed
Speed is useful only when accuracy is preserved. The router determines module placement, so a calm, accurate pass here is generally a better trade-off than finishing with time to spare but more errors.
Spend the extra seconds when you're uncertain
If a router item is genuinely ambiguous, taking a few extra seconds to re-read the stem or stimulus is usually worth it. Mark obvious eliminations first, then commit to your best remaining option.
Don't leave router items blank
There is no penalty for wrong answers. An informed guess is always better than a blank — especially in the router, which feeds every later routing decision.
Stay within the time budget
The router has its own time allotment (~18–21 min in Reading; ~18 min in Listening). Don't over-invest in a single item to the point where you cannot finish the module. Aim for steady pacing rather than perfection on every item.
Module-Stage Strategies: Upper or Lower Module
The upper module typically contains more demanding items — more abstract vocabulary, denser inference, more complex stimuli. The fact that you reached this module suggests your router performance was strong; treat the upper module as an opportunity to demonstrate that ability fully.
Don't panic at the difficulty step
Items in the upper module are designed to be harder. Unfamiliar vocabulary or layered reasoning does not mean you are failing — it is the expected difficulty profile for this path. Stay deliberate.
Inference items reward re-reading
When the answer depends on what is implied rather than stated, go back to the relevant portion of the passage or audio. Don't try to answer from memory under pressure.
Pace yourself across items, not within them
Set rough mental time targets per item and stick to them. Spending too long on a single tough item often costs you on later, more answerable ones.
Placement in the lower module reflects how the router scored you against an undisclosed threshold. ETS has not published how module placement affects the final 1–6 band, so it would be premature to assume any specific score ceiling. The best response is to maximise accuracy here.
Aim for high accuracy on every item
Items in the lower module should be more approachable for you, on average. Strong performance here gives the score model the cleanest possible signal of your ability.
Don't second-guess confident answers
Switching answers on the basis of doubt rather than new information tends to hurt accuracy. Trust your first read when you're confident, and move on.
Treat the experience as diagnostic
After the test, reflect on which router-stage item types felt hardest — vocabulary in context, inference, organisational structure — and focus future preparation there.
Reading Score Maximization Strategies
These strategies apply across both the router and the module, and align with the skills the blueprint identifies as being assessed in Reading.
Build inference and implication skills
Main-idea and detail questions are the foundation; inference questions — where the answer is implied rather than stated — are typical of stronger Reading items. Practice identifying what a passage implies without stating it directly.
Develop active skimming for structure
Before answering questions on academic passages, skim to identify the organisational structure: what is the main argument, where does evidence appear, where does the author make evaluations? This speeds up targeted re-reading during the question phase.
Expand academic vocabulary systematically
Complete the Words items and inference questions both reward contextual vocabulary knowledge. Regular reading in both daily-life and academic registers builds the breadth that short-term review cannot replicate.
Practice timed conditions
Accuracy under time pressure is a skill distinct from accuracy with unlimited time. Regular timed practice sessions train you to make accurate decisions quickly.
Review errors by question type
Track whether your errors come from main idea, detail, inference, vocabulary, or rhetorical-purpose questions. Knowing your error pattern directs your preparation more efficiently than reviewing all question types equally.
Listening Score Maximization Strategies
Listening rewards active engagement with audio — not passive hearing. These strategies address what the blueprint identifies as core listening skills across both modules.
Take structured notes on every audio
Inference, speaker-purpose, and organisational questions are easier to answer from structured notes than memory. Develop a consistent format: main point, key details, speaker stance, signal words.
Listen for speaker intent, not just content
Many listening items ask about why a speaker said something, what they implied, or what their attitude is — not just what was said. Train yourself to track tone, emphasis, and purpose as you listen.
Build stamina through extended listening practice
Academic lectures and conversations can be dense. Regular exposure to authentic academic audio — podcasts, lectures, structured conversations — builds the concentration and processing speed both modules reward.
Practice Listen and Choose a Response carefully
This task tests communicative intent — what is the appropriate next reply to what was just said? Distinctions between options can be subtle. Analyse why each wrong option is wrong, not just why the right one is right.
Train inference listening
Practice with materials that include implied meaning, hedging, qualification, and idiomatic expressions — the features of natural academic and conversational speech that listening items frequently probe.
Common MST Myths
Misconceptions about adaptive testing lead to poor strategy decisions. These are the most common myths about TOEFL MST.
Myth: A harder upper module means you're doing badly
✓ Encountering more challenging items in the upper module is generally interpreted as a sign of strong router performance. It does not mean you are failing — it is simply the difficulty profile ETS has matched to that path.
Myth: You can make up for a weak router with a perfect upper module
✓ The blueprint confirms a staged structure where the router precedes the lower or upper module. ETS has not detailed exactly how router and module responses combine, but the staged design suggests that earlier-stage performance plays a meaningful role in subsequent routing and scoring.
Myth: Hard items are explicitly worth more raw points than easy items
✓ ETS has not publicly disclosed how individual item difficulty translates into the final 1–6 band score. Modern adaptive psychometric models do typically weight responses by item characteristics, but the precise weighting for TOEFL 2026 has not been published.
Myth: The lower module makes a top band impossible
✓ The blueprint does not state that placement in the lower module forecloses any particular score outcome. Until ETS publishes the scoring algorithm, treat module placement as an indicator of router performance — not a definitive cap.
Myth: The adaptive system penalizes slow test-takers
✓ Reading and Listening have overall section time limits, not per-item timers. Within the time budget, spending extra seconds on a router item to confirm accuracy is a reasonable trade-off.
Myth: Writing and Speaking also adapt to your performance
✓ The blueprint is explicit: Writing and Speaking are linear. Every test-taker of a specific form receives the same set of tasks. Adaptive logic applies only to Reading and Listening.
Practice TOEFL Reading and Listening with Adaptive Difficulty
LingoLeap's practice environment reflects the adaptive structure of TOEFL 2026 Reading and Listening — so your preparation builds the skills that matter under real testing conditions.
Start Practicing Reading and ListeningFrequently Asked Questions
How many stages are in TOEFL 2026 Reading and Listening?⌄
The ETS 2026 blueprint confirms that Reading and Listening each use a two-stage adaptive structure: a router module followed by either a lower or upper module. All test-takers begin with the same router. Performance in the router appears to influence which module test-takers see next, though ETS has not disclosed the exact routing thresholds.
Can I tell during the test which stage I'm in?⌄
No. ETS does not notify you when a stage boundary occurs or which module you have been routed to. The most reliable strategy is consistent accuracy throughout — particularly in the router portion, which the blueprint identifies as the entry stage shared by every test-taker.
Does the upper module always produce a higher score than the lower module?⌄
Not necessarily. The ETS blueprint confirms the staged adaptive structure but does not publicly disclose how router and module performance combine into the final 1–6 band score. Different module difficulty paths likely interact with score calculation, though the exact mechanics remain undisclosed. The blueprint also confirms that some items in Reading and Listening are unscored (used to calibrate future items), which adds another layer that ETS has not detailed.
Do MST strategies apply to TOEFL Writing and Speaking?⌄
No. The blueprint confirms Writing and Speaking are linear — all test-takers of a specific form receive the same set of tasks. Adaptive (router/module) logic applies only to Reading and Listening.
Should I guess on router-stage questions I don't know?⌄
Yes. There is no penalty for wrong answers in TOEFL 2026. If you are uncertain, make your best informed guess rather than leaving the question blank. The router is the only stage every test-taker shares, so attempting every item is a low-risk choice.
Are there unscored questions in TOEFL 2026 Reading and Listening?⌄
Yes. The blueprint explicitly states that the item counts for Reading and Listening include non-scored items used to gather data on item performance for future test development. This is not entirely new — pre-2026 TOEFL versions also included experimental items — but the 2026 blueprint now documents it more transparently. Test-takers cannot identify which items are unscored, so the practical advice remains: answer every question carefully.
Related Guides
TOEFL Reading 2026
Complete guide to the Reading section tasks and strategies.
Read guide →
TOEFL Listening 2026
Complete guide to the Listening section tasks and strategies.
Read guide →
TOEFL 2026 Format and Scoring
How all four TOEFL 2026 sections are structured and scored.
Read guide →
TOEFL 2026 Score Scale
What each band level means across sections.
Read guide →
How to Score a 6 in Speaking
Rubric strategies for the Speaking section's top band.
Read guide →
How to Score a 6 in Writing
Rubric strategies for the Writing section's top band.
Read guide →
TOEFL Score Hub
All TOEFL score tools and guides in one place.
Read guide →
About this guide
Written by the LingoLeap Research Team — TOEFL preparation specialists who analyse ETS published documentation, including the April 2026 TOEFL iBT Test: 2026 Update — Test Blueprint and Specifications Document. Structural details (router/module design, item counts, timings, AI-scored task identification) reflect that blueprint. Where ETS has not publicly disclosed scoring or routing mechanics, this guide notes the uncertainty rather than speculating. Last reviewed May 2026.