The EQAO assessment, taken apart

Start here

Five facts that put the rest in proportion

Every one of these is EQAO's own position, not a reassurance invented for the occasion. Read them before anything else on this page.

How to use this page

This is the assessment companion to the Grade 3 curriculum guide. The curriculum guide tells you what a child is meant to learn; this one tells you what EQAO actually puts in front of them in May, how it is scored, and what the province's own data says about where children come unstuck.

Who this is for. Teachers planning the year, and families who want to understand what is coming. Use the Teachers / Families switch at the top to hide the sections aimed at the other audience, and the search box to jump straight to a topic.

So what is this guide arguing?

The best preparation for EQAO is the ordinary, well-taught work of the year, and EQAO says so itself. Nothing here is a case for drilling. The argument is narrower and, I think, defensible:

The four levels

Reading, writing and mathematics are reported independently. A child can sit at Level 3 in reading and Level 2 in writing, and that is ordinary.

Logistics

Key dates

Schools choose their own dates inside the window. Grade 3 and Grade 6 need not sit at the same time, and a teacher may run the assessment with small groups — or one student at a time — on different days.

Part one

The shape of the assessment

Two components, administered separately. They are built on completely different principles, and this is the single most misunderstood thing about the assessment.

Timing

These are the suggested times from the administration user guide. Additional time may be given to all students — it is not an accommodation, it is built into the design.

Roughly four hours and ten minutes of suggested time in total, spread across eight sittings. Session A is shorter than the rest.

Order. Sessions must run A → B → C → D, and stages 1 → 2 → 3 → 4. But the two components can be interleaved however the teacher likes: all four language sessions then all four maths stages, or alternating A, 1, B, 2, C, 3, D, 4.

The adaptive design — and why it worries people more than it should

Mathematics uses a multi-stage computer adaptive model. Every child starts Stage 1 on a module of medium overall difficulty. Based on how they do, Stage 2 serves either a low/medium or a medium/high module. The same routing happens again between Stage 3 and Stage 4.

Three things follow, and all three are reassuring.

1. Every child answers the same number of questions from each strand. The blueprint is fixed. Routing changes difficulty, not coverage.
2. Outcomes are placed on a single scale regardless of route. EQAO's position, stated directly: a student's outcome accounts for their responses to all operational questions and the difficulty of each of those questions, regardless of how the student is routed, and overall levels of achievement are comparable across routes.
3. Field-test questions do not affect routing. The four embedded trial questions are excluded both from routing decisions and from the final result.

Language, by contrast, is linear — several equivalent forms built to the same specification, assembled before administration and fixed in length. Nothing adapts.

The platform, and its tools

Fully digital, on school-provided devices. Available to every student, without any accommodation paperwork:

Rough notes on screen are a junior-division feature. Grade 3 students use paper for rough work.

Permitted and not permitted

Navigation rules worth knowing

• Students can move backwards and change answers freely — right up until they submit that session or stage.
• Once a session or stage is submitted, it is closed. No returning to it.
• Any unsubmitted session is automatically submitted at 5 p.m. daily.
• Students using paper response sheets for the writing open-response must finish writing before the session is submitted — after submission, the question is no longer visible.

Part two

What is actually assessed

The assessment is built from an official blueprint that maps every question to a curriculum expectation. It is public, and unusually specific. What follows is that blueprint, in plain language.

Reading

Three skills, stated plainly in the Framework:

1. Understanding explicitly stated information and ideas — it is in the text, findable
2. Understanding implicitly stated information and ideas — inference
3. Responding to a reading selection by making connections, bringing the reader's own knowledge and experience to the text

That is the whole of it. Not fluency, not decoding speed, not book knowledge. Two texts, and their lengths are specified — which is useful to know.

Reading score points — 27 in total

Note that 14 selected-response questions yield 19 points. Some are worth more than one — the multi-part ones, where a child drags three items into place and can be partly right. Partial credit exists, so attempting all parts of a multi-part question is materially better than attempting one.

Which expectations are in play

From the 2023 Language curriculum. Strand A is not assessed at all. Neither is B1, oral and non-verbal communication.

The open-response reading questions, decoded

Two of them, one per text, both worth four points, scored by trained educators against a public rubric. The gap between a 2 and a 3 is not about how much a child writes.

Part two, continued

Writing

The weakest of the three components province-wide, and for years. It is also where preparation has the clearest legitimate role, because the criteria are published and narrow.

Three skills:

1. Developing a main idea with sufficient supporting details
2. Organising information and ideas coherently
3. Using conventions — syntax and sentence structure, grammar, cohesive ties, capitalisation and punctuation — in a way that does not distract from clear communication

Read the third carefully. The standard is not correctness. It is not getting in the way.

The single writing task is worth 7 of 20 points. It is the highest-value item in the entire language component.

Those 7 points come from two rubrics applied to the same piece of writing — a 4-point Topic Development scale and a 3-point Conventions scale. The Framework states the 7-point total and publishes both rubrics; the 4-plus-3 split is the arithmetic that follows, not a figure EQAO states in those words.

How that one task is scored — twice

Topic Development — four points

Conventions — three points, on a completely different scale

Which expectations are in play

Strand B — the same B2 and B3 expectations as reading. Conventions are assessed in both the selected-response questions and inside the open-response answer.

Part two, continued

Mathematics

Forty operational questions. The proportions are fixed, and every student answers the same number from each strand regardless of how they are routed.

Financial Literacy is folded into Number because it has only one specific expectation at this grade — F1.1, calculating change.

What is not assessed

• Strand A — Social-Emotional Learning skills and the Mathematical Processes. Students may need to use the processes, but no question is mapped to them.
• C4 Mathematical Modelling — there are currently no EQAO questions mapped to this overall expectation, and no specific expectations under it.
• Communication — of the four achievement chart categories, mathematics questions map to only three. There is no open-response mathematics question at Grade 3 and nothing to explain in words.

The three categories

Every question is mapped to one of these. All three appear in every stage.

Part two, continued

Mathematics strand by strand

Every specific expectation EQAO's blueprint maps questions to, in curriculum order. Codes follow the 2020 Mathematics curriculum.

B. Number — B1 Number Sense

B. Number — B2 Operations

F. Financial Literacy

C. Algebra

D. Data

E. Spatial Sense — E1 Geometric and Spatial Reasoning

E. Spatial Sense — E2 Measurement

Part three

Where marks are actually lost

This is the part that cannot be guessed at, and the part most preparation materials get wrong. Two separate EQAO publications underpin it, and it is worth keeping them apart.

EQAO's four themes

1 · Proportional reasoning — the big one

What Level 2 students could do: add and subtract accurately when told explicitly to do so.

Reaching for addition when the situation requires multiplication. Asked how many bananas are in a box holding four times as many as six apples, students at Level 2 tended to choose ten. They correctly recognised that the answer should grow — and then added. The mirror image appeared in division: asked how many groups of five are in 45, they chose 40. They recognised a reduction, and subtracted.

Scales on number lines. Given a number line whose tick marks advanced in fives, only about a third of Level 2 students identified two labelled points correctly, against over half at Level 3. The failure was not arithmetic — it was not grasping that a number line's scale must be consistent.

Area, answered halfway. Shown a book partly covered with identical square tiles and asked how many more tiles were needed, only a quarter of Level 2 students answered correctly. The dominant wrong answer was the number of tiles already there. They did the first step and stopped.

2 · Algebraic reasoning

What Level 2 students could do: continue a pattern that skip-counts by tens.

Equivalence. Asked which expression equals 4 × 2, students in difficulty chose 4 ÷ 2 — inverse operations conflated. Asked for an expression equal to a number such as 345, they gravitated to an addition option even when the correct answer involved subtraction.

Missing digits. Given a subtraction with two digits blanked out, about a third of Level 2 students found them, against over half at Level 3. The characteristic error was subtracting where regrouping required addition.

Shrinking patterns, and applying the rule to every term. Given a pattern decreasing by six each time, roughly half of Level 2 students found it, against more than 80% at Level 3. The revealing detail: they tended to pick a sequence whose second number was right and whose later numbers drifted. They applied the rule once and then estimated.

Pictograph keys. Asked to read a pictograph where each symbol stood for four items, about 40% of Level 2 students answered correctly against 85% at Level 3. The wrong answer was counting the symbols — one-to-one correspondence, ignoring the key.

3 · Computational thinking

Movement on a grid. Asked to describe a route involving forward moves and a turn, Level 2 students made two errors, often together: counting the starting square as a move, and swapping clockwise and counterclockwise.

Calculating change. Only a third of Level 2 students handled a change question correctly, against three quarters at Level 3. The structure defeats them: subtract to find the change, then add up a set of coins to find which option matches. Two operations, in sequence, with a representation change in between.

4 · Mathematical language

What Level 2 students could do: use certain and impossible correctly.

Likely and unlikely. About 55% of Level 2 students against 85% at Level 3. The misconception is precise and worth naming: they treat “likely” as meaning “possible”. Anything that could happen is likely. The idea of a continuum from impossible to certain, with degrees in between, has not landed.

Coding vocabulary. Asked which line of code was a loop, or which would repeat an instruction a set number of times, Level 2 students struggled — the terminology, not the logic.

Faces, edges and vertices. About half of Level 2 students could not correctly count them on a given 3-D object.

Mode. Asked for the mode from a frequency table, they chose the largest value rather than the most frequent one.

The hardest released questions, by the numbers

From the November 2025 released set. Bars show the share of students at each level who answered correctly. Several are questions most people would assume are easy.

And the ones that were easy

Worth knowing too, so that time is not spent where it is not needed. At Level 3, over 85% answered correctly on:

Part four

The provincial picture

Where the province sits, from the 2024–25 results for students in English-language schools. Useful for calibration: it tells you what “normal” looks like before you decide anything is wrong.

Students with special education needs

Writing is the one component moving the wrong way for this group, and by a widening margin against the provincial figure.

English language learners

What students say about themselves

From the Student Questionnaire, which 97% of participating students completed. Worth reading beside the achievement data.

On mindset: 89% of students believe a person can always get better at mathematics, and 73% believe almost everyone can understand mathematics if they work at it. 79% say they keep trying after a mistake.

On technology: 33% of students indicated they are able to use the internet at home to complete school work, and 54% indicated they use technology to learn new things.

Part five

Access, accommodations and exemptions

Mainly for teachers and administrators, but families should know these exist.

Available to every student, no paperwork

Requiring an IEP or a documented special consideration

• Alternative Version formats — Braille, e-Braille, printable PDF in large or regular print, online with or without full descriptions of mathematics graphics (single-selection questions only)
• Assistive technology — Read&Write for Chrome, Kurzweil, NVDA, VoiceOver, Read Aloud, JAWS, ChromeVox, Natural Reader
• Voice-to-text
• Verbatim scribing
• Sign language or oral interpreter
• Extended periodic supervised breaks

Special considerations that qualify a student without an IEP: new to school (with supervisory officer permission) · temporary or unique circumstances (with school administrator permission) · English language learner in the early stages of acquisition.

Scribing, briefly

Scribes must be adults and must not be relatives of the student. They transcribe exactly what is dictated, assume each sentence starts with a capital and ends with a period, read it back if asked, show the student the transcription and correct only as directed. They must not clarify, explain, translate, prompt, react, or suggest strategies.

Exemptions

Decided before the assessment, for each student individually, in consultation with the student, parents or guardians and teaching staff — and with the consent of the parents or guardians.

Two are mandatory rather than discretionary:

• A student must be exempted from reading if built-in or external text-to-speech is not sufficient to give them access to the reading assessment.
• A student must be exempted from mathematics if they require mathematics terms to be defined for them.

Part six

The checklists

Everything above, converted into things you can actually tick off. Seven lists, 221 items. Items marked ⚠ are ones EQAO's own data flags as where students at Level 2 come apart — filter to those if time is short.

Part seven

If you only do twelve things

In priority order, weighted by how much of the assessment they touch and how reliably the evidence says students lose marks there.

1. Multiplicative thinking, taught as distinct from additive thinking

   The most diagnostic error in EQAO's entire analysis. Children recognise the direction of a problem and pick the wrong operation. Sorting mixed word problems into two piles, with justification, is the intervention — and it costs fifteen minutes a week.

2. Multi-step questions, named as such

   “How many steps does this need?” before starting, and “did I do them all?” after. The area question answered halfway, the change question stopped after the subtraction, the two-purchase problem — all the same failure.

3. Instant recall of ×2, ×5, ×10 and the related division facts

   B2.2, and the year's highest-value single outcome. Not all the tables — those are Grade 4.

4. Evidence plus “which shows that…” in every reading response

   The published rubric turns on whether the child connects the evidence to the point or leaves the reader to do it. One taught sentence structure moves a 20 to a 30.

5. Precise mathematical vocabulary, on a wall, all year

   Faces, edges, vertices. Impossible, unlikely, equally likely, likely, certain. Loop, repeat, sequence. Mode, mean, frequency. Perimeter, area. EQAO found language failures across every strand.

6. Pattern rules applied to every term, and gaps counted rather than numbers

   The hardest released question in the 2025 set, failed by most students who met the standard.

7. Writing that finishes

   Organisation is scored. A piece with no ending is disorganised by definition. Plan for three developed paragraphs, not five started ones.

8. Scales and keys, always read first

   Number lines, bar graphs, pictographs. Pointing at the key before answering, every single time, until it is automatic.

9. All three writing forms — narrative, persuasive, informational

   The prompt could be any of them. A child who has only written stories is exposed.

10. Faces, edges and vertices, counted on real objects

    Half of Level 2 students could not do this. It is twenty minutes with a box, a tin and a paper pyramid.

11. The sample test, on the real devices

    Free, official, and it removes an entire category of lost marks that has nothing to do with what the child knows.

12. Saying, clearly and more than once, that there is enough time

    Extra time may be given to all students. Most children do not know this, and a surprising number of the marks lost to rushing are lost to a belief that is simply false.

Reference

Common mistakes in preparation

Before you rely on this

Verification status

This section exists so reviewers know what has been checked, against what, and where the soft spots are. Concentrate review effort on the confidence column.

What each claim rests on

The four things a reviewer should actively check

1 · Session D. EQAO's own materials describe the language component inconsistently — see the note in Part One. If anyone on the team has a definitive answer from a board assessment lead, it belongs in the guide.

2 · The fraction codes — a trap worth knowing about. This guide follows EQAO's blueprint, which lists Grade 3 Number Sense as B1.1 to B1.7, with no B1.8, and Operations as B2.1 to B2.9. Both were independently confirmed against the Ministry-funded GEEM resource.

3 · The 33% technology figure. Flagged in Part Four. Reported accurately; probably should not be leaned on.

4 · Anything date-stamped. The 2027 window, the framework edition and the user guide were all current at the time of writing. EQAO revises annually. Check before distribution.

What is deliberately not here

• No reproduced assessment questions. Question types and error patterns are described; no stem, option set or passage is copied.
• No practice questions. This is a guide to the assessment, not a practice book. Practice material aligned to the real thing already exists free from EQAO in the released questions and the sample test.
• No predictions. Nothing here forecasts what will be on the 2027 assessment. The blueprint is fixed; the content is not knowable.
• No claim that preparation raises scores. The argument made is narrower: that a small set of avoidable, non-academic losses can be removed, and that teaching the curriculum well does the rest.

Reference

Sources and copyright

Everything here is a paraphrase in plain language. No EQAO or curriculum text is reproduced verbatim, no assessment questions are reproduced, and the rubrics are described rather than copied. EQAO materials and the Ontario curriculum documents are Crown copyright — © King's Printer for Ontario. Expectation codes are given so you can find the original wording, which is always the authority.

Official sources