Every TOEFL Writing response — whether Write an Email or Write for an Academic Discussion — is scored against four consistent criteria: task completion, organization, grammar, and vocabulary. Build a Sentence is scored differently, since it's a more mechanical, single-correct-answer task rather than an open-ended writing response. Understanding exactly what separates a top-scoring response from a middling one on each criterion is the fastest way to turn generic "write better" advice into specific, practicable technique.
This guide breaks down the rubric for both open-ended Writing task types in detail. If you haven't yet reviewed the section's overall structure, start with our TOEFL Writing section format guide first.
Take the free TOEFL Benchmark Test and get your estimated band score in about 45 minutes.
As with Speaking, a response doesn't need to be flawless across all four criteria to score well, but a response that's notably weak in any one area is capped by that weakness — a grammatically flawless response that doesn't fully address the prompt scores lower than test-takers sometimes expect, because task completion is being evaluated just as seriously as accuracy.
This task asks you to respond to a simulated class discussion thread, engaging with the professor's question and other students' posted viewpoints. Here's what separates each score level.
| Score Level | What It Requires |
|---|---|
| Top score | Direct relevance to the discussion; well-developed ideas with specific details and examples; strong language control with real sentence variety; high clarity and smooth cohesion; very few errors. |
| High-mid score (~80%) | Clearly addresses the topic with adequate elaboration; good sentence and vocabulary variety; few errors, none that affect meaning. |
| Mid score (~60%) | Main idea is present but elaboration is uneven; limited sentence and vocabulary variety; noticeable grammar errors that reduce smoothness. |
| Low-mid score (~40%) | Ideas are poorly developed or only partly relevant; limited vocabulary and sentence structures; many grammar errors that interfere with understanding. |
| Low score (~20%) | Ineffective attempt with few coherent ideas; severely limited vocabulary and structure; frequent serious errors with minimal original language. |
| No credit (0) | Blank response, non-English, copying the prompt directly, rejecting the topic, or a completely unrelated response. |
The gap between a top score and a mid-range score is rarely about a single dramatic error — it's usually the combined effect of thinner elaboration, less varied sentence structure, and a higher error rate, each individually modest but compounding together. See our Academic Discussion task guide for the specific response structure this task rewards, and our sample essays at Band 3, 5, and 6 to see these differences applied to real responses.
This task asks you to write an email responding to a given scenario, and adds one criterion the Academic Discussion task doesn't emphasize the same way: appropriate tone and format for an email specifically, not just content quality.
| Score Level | What It Requires |
|---|---|
| Top score | Fully addresses all parts of the prompt; appropriate tone and format for an email; clear organization; excellent grammar and vocabulary. |
| High-mid score (~80%) | Addresses main points with minor omissions; generally appropriate tone; good organization; few errors. |
| Mid score (~60%) | Addresses the topic but misses some requirements; inconsistent tone; adequate organization; noticeable errors. |
| Low-mid score (~40%) | Partially addresses the topic; inappropriate tone or format issues; weak organization; frequent errors. |
| Low score (~20%) | Barely addresses the topic; major format or tone issues; poor organization; many serious errors. |
| No credit (0) | No response, not in English, or completely off-topic. |
The tone and format criterion is what most distinguishes this task from Academic Discussion. A response with strong content but a tone that's too casual for a formal request, or too formal for a casual message to a friend, loses points here even if the underlying grammar and vocabulary are strong. See our Write an Email task guide for exactly how to calibrate tone to different prompt scenarios.
Take the free TOEFL Benchmark Test and get your estimated band score in about 45 minutes.
Build a Sentence is scored differently from the two open-ended tasks covered in detail above — it's a more mechanical task with a defined correct construction, testing your ability to accurately arrange sentence components rather than producing original, developed writing. There's no task-completion, organization, or vocabulary-range criterion in the same sense, since the task itself constrains what you're producing. Instead, scoring centers on whether your constructed sentence is grammatically correct and matches an accepted construction. See our Build a Sentence task guide for the specific technique this task rewards.
Because your overall Writing band combines performance across all three task types, a strong Build a Sentence performance — which tends to reward careful, mechanical accuracy rather than creative writing ability — can meaningfully support test-takers who find the two open-ended tasks more challenging, and is worth practicing deliberately rather than treating as a minor warm-up task.
Your overall Writing section band, on the 1.0–6.0 scale described in our good TOEFL score guide, reflects your combined performance across Build a Sentence, Write an Email, and Write for an Academic Discussion — not any single response in isolation. This means a single underdeveloped response, perhaps from an unfamiliar or challenging prompt, is unlikely by itself to define your entire Writing band, but a consistent pattern of weakness on the same criterion — say, task completion, across multiple practice responses — will show up clearly and is worth addressing directly rather than dismissing as one-off bad luck.
Task completion is often the criterion test-takers most underestimate, because it's not about writing quality in the usual sense — it's about whether you actually answered everything the prompt asked, in full. Two common ways this criterion gets lost:
Before writing, it's worth explicitly listing every distinct thing the prompt is asking for, then checking your finished response against that list — a simple habit that catches task-completion gaps before they cost you points.
Start structured TOEFL preparation built around the current 2026 format.
Read the prompt twice before writing, and identify every distinct requirement explicitly — don't rely on a single read-through and general impression of what's being asked. For Write for an Academic Discussion, directly reference at least one other posted viewpoint by name or idea, rather than only stating your own opinion in isolation.
Use a clear, predictable structure: a direct opening statement, one or two well-developed supporting points each in their own paragraph or clearly separated section, and a brief closing. Use explicit transition words and phrases between ideas — "in addition," "however," "as a result" — since these are exactly what graders and readers use to judge whether a response flows logically rather than reading as a disconnected list of points.
Vary your sentence structure deliberately, mixing simple, compound, and complex sentences, since sentence variety is explicitly rewarded at the higher score levels — but only if used accurately. A single well-formed complex sentence demonstrates more control than several grammatically safe but simple sentences strung together. Proofread specifically for the error types you know you make repeatedly, rather than reading generically for "mistakes."
Replace generic, overused words ("good," "bad," "important," "a lot of") with more precise alternatives appropriate to the topic, and avoid repeating the same key word multiple times in a short response when a synonym would demonstrate more range. See our TOEFL vocabulary guide for building the underlying word bank this draws from.
Imagine an Academic Discussion prompt asking whether university students should be required to complete an internship before graduating.
"I think internships are good for students. They learn a lot about the real world. Also school is very theoretical and internship give practical skill. So I agree students should do internship before graduating." This response addresses the topic and states a clear position, but elaboration is thin — one general reason without a specific example — sentence structures are simple and repetitive, and there are noticeable grammar errors ("internship give" instead of "internships give"). This reflects the mid-range level of the rubric: main idea present, but uneven elaboration and noticeable errors.
"I agree that a required internship makes sense, mainly because it closes the gap between what students learn in theory and what employers actually expect on day one. For example, a computer science student might understand algorithms deeply in the classroom but have never worked with the messy, undocumented codebases typical of a real workplace — an internship forces that adjustment before it matters for a full-time job. That said, I'd add one consideration Maria raised in her post: not every field has abundant internship placements, so any requirement would need flexible alternatives for students in less internship-heavy majors." This response directly engages the prompt, develops a specific example, references another student's post by name (directly satisfying the Academic Discussion task's core requirement), uses varied sentence structure including a complex sentence with a dash-set-off clause, and maintains accurate grammar throughout. This is the level of engagement and elaboration the top score level describes.
Take the free TOEFL Benchmark Test and get your estimated band score in about 45 minutes.
Once you understand the four criteria, review your own practice responses against each one separately rather than judging a response as generically "good" or "not good." After writing a practice response, check it explicitly: Did I address every part of the prompt? Is my structure clear with logical connections between ideas? Is my grammar accurate across varied sentence types? Is my vocabulary precise and varied rather than repetitive? This structured self-review mirrors how an AI grader or human examiner actually evaluates a response, and — combined with real feedback rather than self-assessment alone, since self-assessment is notoriously unreliable for judging your own writing objectively — is the most direct way to turn understanding the rubric into an actual score improvement.
AI-graded feedback on Write an Email and Write for an Academic Discussion, broken down by criterion.