The short answer: a GRE text completion question is a passage of one to five sentences with one to three blanks. A one-blank item has five choices; a two- or three-blank item has three choices per blank, and ETS gives no credit for partially correct answers, so every blank must be right. The method that works is to read for the sentence's logic, predict each blank in your own words, then match and eliminate.
Text completion and its sister type, sentence equivalence, are the GRE Verbal questions that lean hardest on vocabulary. ETS says about half of the Verbal measure is reading comprehension, so these sentence-based items make up roughly the other half. Most test takers treat them as pure vocabulary quizzes: know the word, get the point; don't know it, guess. That framing is wrong, and it costs points.
Text completion questions are logic problems with vocabulary as the medium. The correct answer is set by the sentence's internal structure (its direction, its signal words, its argument), not by which word sounds impressive or which one you recognize.
That means these questions are winnable even when your vocabulary has gaps, because the sentence's logic narrows the answer before you look at the choices. This guide gives you a complete, step-by-step system for every blank. Its sister item type, sentence equivalence, uses the same prediction step but asks for two choices that produce the same meaning, and both depend on how the GRE actually tests vocabulary.
Before diving in, get a baseline. Twenty timed text completion practice questions will show whether your misses come from logic, vocabulary or time, and that makes this guide far more actionable.
The Anatomy of a Text Completion Question
Per ETS's Verbal Reasoning page (checked October 2026), each text completion passage is one to five sentences long with one to three blanks. Choices are chosen independently for each blank, and exactly one choice per blank is correct.
| Form | Blanks | Choices | Odds of a blind guess |
|---|---|---|---|
| One-blank | 1 | 5 choices | 1 in 5 |
| Two-blank | 2 | 3 choices per blank, chosen independently | 1 in 9 |
| Three-blank | 3 | 3 choices per blank, chosen independently | 1 in 27 |
The sentence always carries a logical argument: a claim, a contrast, a cause and effect, or a qualification, and the blanks fall at the critical junctions of that argument. The correct answer for each blank is the one that satisfies the sentence's logic and reads coherently in grammar and style.
There are no trick answers in the sense of wordplay. Every wrong answer is wrong because it contradicts or fails to fit the sentence's logic, not because of an outside fact about the world.
Quick GRE check
3 questions, about 2 minutes. How would you do on the real test?
Verbal · Sentence Equivalence
The new intern was so blank that she completed in two days a backlog her predecessor had let pile up for weeks.
Time Targets Per Format
ETS scores every text completion question all-or-nothing: there is no credit for partially correct answers, so a three-blank item with two blanks right earns the same as one with none right. Time is the other constraint. These are the targets we recommend:
| Question type | Time target | Why |
|---|---|---|
| One-blank text completion | 45 seconds | The sentence is short, the logic is usually clear, and a good prediction narrows five choices quickly. |
| Two-blank text completion | 90 seconds | Two decisions that interact, which needs one extra pass through the sentence. |
| Three-blank text completion | 120 seconds | Three decisions, a longer passage, and the risk of cascade errors. Manage your time here. |
These are targets, not hard stops. A question that genuinely stumps you should get a provisional answer and a mark. Three minutes on one item is almost always worse than banking a correct answer elsewhere in the section.
Where Text Completion Fits in the Current Verbal Sections
Since September 2023 the GRE has two Verbal sections: 12 questions in 18 minutes, then 15 questions in 23 minutes (ETS test structure page). ETS does not publish an exact count for each question type. It says about half of the Verbal measure is reading comprehension. Magoosh's breakdown of the shorter test (checked October 2026) estimates about 7 text completion, 7 sentence equivalence and 13 reading comprehension questions across the two sections, which works out to roughly three or four text completion items per section.
| Verbal section | Questions and time | Likely text completion items | Time they should take |
|---|---|---|---|
| Section 1 | 12 in 18 minutes | About 3 to 4 | About 4 to 5 minutes, leaving 13 or more for passages and sentence equivalence |
| Section 2 | 15 in 23 minutes | About 3 to 4 | About 4 to 5 minutes, leaving about 18 for the rest |
Two things follow. First, text completion is a small number of questions, but because the Verbal measure has only 27 questions in total, each one is a meaningful share of your raw score. Second, the first Verbal section also decides the difficulty of the second, so careless misses on early one-blank items are expensive twice. Our guide to how the GRE's section-adaptive scoring works explains the routing.
Step 1: Read the Sentence for Its Spine, Not Its Blanks
The most important habit in text completion: before you read the answer choices, read the sentence for its spine. The spine is the logical argument the sentence makes, stripped of the blanks and the decoration, reduced to its core claim and its direction.
The tools for finding the spine are signal words. ETS's own advice is to look for significant words or phrases, the structural words that show how parts of the passage relate. They fall into two main categories:
| Signal type | Effect on the blank | Common examples |
|---|---|---|
| Continuation | The blank carries the SAME direction as the rest of the sentence: positive stays positive, negative stays negative. The blank amplifies or extends the surrounding logic. | moreover, furthermore, indeed, consequently, therefore, thus, in fact, as a result, additionally, accordingly |
| Contrast | The blank carries the OPPOSITE direction from the rest of the sentence. The blank inverts the surrounding logic. | however, despite, although, yet, but, while, whereas, even though, paradoxically, in contrast, nonetheless |
These signals change the correct answer completely. Consider two sentences identical except for the signal word:
- "The mayor was known for being _______, and indeed her handling of the budget crisis confirmed that reputation." 'Indeed' is a continuation signal. The blank must match whatever quality her budget handling confirmed, so it is something like 'decisive', 'shrewd' or 'pragmatic'.
- "Although the mayor was known for being _______, her handling of the budget crisis surprised observers." 'Although' is a contrast signal. The blank must carry the opposite of what surprised observers. If they were surprised by decisive action, the blank is something like 'indecisive' or 'tentative'.
Identical structure, opposite answers, because of one signal word. That is why reading the spine before the choices is the most important step in the system.
Note the signal words on your scratch paper as you read (at home, on your whiteboard). It takes two seconds and prevents the most common error: the right kind of word in the wrong direction.
Step 2: Predict Before Looking at Choices
Once you have the spine and the signal words, cover the answer choices and write down (or say to yourself) your own word for each blank. This is also ETS's own advice: try to fill the blanks with words that seem appropriate, then look for them among the choices.
Prediction works because it makes you use the sentence's logic rather than the choices' appeal. Read the choices first and you get anchored by words that sound plausible or familiar, even when they contradict the sentence. Predict first and you have a target to test each choice against.
Your prediction does not need to be a specific word. It only needs to capture the direction. Any of these is enough:
- 'Something positive that means capable.'
- 'Something that means criticism or attack.'
- 'A word describing reluctance or hesitation.'
The more specific your prediction, the faster you will match it, but even a directional prediction usually eliminates half the choices at once.
Prediction on a Two-Blank Question
Here is prediction on a full two-blank question. "The documentary was praised for its (i) _______ portrayal of the refugee crisis, which critics found far more (ii) _______ than the sensationalized coverage common on cable news." Blank (i): sanitized, unflinching, lurid. Blank (ii): illuminating, misleading, sensational.
- Read the spine. The sentence praises the documentary, and the second clause compares it favorably with 'sensationalized coverage'. Blank (ii) must be a virtue that sensational news lacks: predict 'informative' or 'insightful'.
- Blank (i) describes the portrayal that earned the praise, and it must not be sensational, because the sentence sets the film against sensationalism. Predict 'honest' or 'clear-eyed'.
- Match. For (ii), 'misleading' is negative and 'sensational' is the very thing the film is contrasted with, so 'illuminating' fits. For (i), 'sanitized' means cleaned up to the point of distortion, which is not praise for a documentary, and 'lurid' means sensational, which contradicts the contrast. 'Unflinching' (honest, not looking away) fits. Answer: unflinching, illuminating.
Predict the blanks in the order that is most constrained by the surrounding text, not necessarily left to right. ETS makes the same point: do not assume the first blank is the one to fill first.
Step 3: Match and Eliminate
With a prediction in hand, test each choice against two criteria: does it match your prediction's direction, and does it fit the sentence's logic when you substitute it in? The direction test is fast and removes broad categories of wrong answers. The substitution test is finer and catches near-synonym traps.
Four elimination rules cover the vast majority of wrong answers. Apply them in order: direction first because it is fastest, then the finer tests on the two or three survivors.
- Opposite direction: the choice carries the opposite logical direction from what the sentence requires. If the sentence calls for a positive quality and the choice is clearly negative (or vice versa), eliminate it immediately. This often removes two of five choices on a one-blank question.
- Too extreme: the choice has the right direction but an intensity the sentence cannot support. A sentence that says a scientist was 'somewhat critical' of a theory cannot support 'contemptuous'. A sentence calling an executive 'regarded as effective' cannot support 'legendary'. Match intensity, not just direction.
- Plausible but not supported: the choice describes something that could be true of the subject in real life but is not what the sentence's logic requires. A sentence about a novelist's prose style is not asking about her personal character.
- Unintended meaning shift: substituted in, the choice changes what the sentence means in a way the context cannot accommodate. This often involves words with more than one sense. 'Critical' can mean important (a critical juncture) or fault-finding (a critical review); if the context points to one sense and the choice mainly carries the other, eliminate it.
Finish with ETS's last check: read the completed passage once more and confirm it is logically, grammatically and stylistically coherent.
Two-Blank Text Completion: The Interaction Problem
In two-blank questions the blanks interact. They are not two mini-questions; they are part of the same sentence's logic, and your choice for one blank constrains what fits the other.
There are two basic approaches, and a hybrid that beats both:
| Approach | How it works |
|---|---|
| Column-by-column | Finds the best answer for each blank independently, without testing whether the combination makes sense. |
| Combined | Tests all nine combinations of a 3 by 3 grid for coherence. ETS itself advises against grinding through every combination. |
| Hybrid (recommended) | Solve the more constrained blank first, then use that answer to narrow the choices for the other blank, and finally read the pair together. |
The more constrained blank is the one with more surrounding context: more signal words nearby and more specific logical requirements.
Example: "The senator's proposal was simultaneously _______ in its ambition and _______ in its details, a combination that left supporters uncertain whether to celebrate or despair." Blank 1 gets a positive word (ambition worth celebrating), something like 'sweeping' or 'visionary'. Blank 2 gets a negative word (details worth despairing over), something like 'vague' or 'deficient'.
- A 'sweeping' proposal with 'vague' details fits: the tension is exactly what the sentence describes.
- A 'modest' proposal with 'vague' details does not, because a modest ambition gives supporters nothing to celebrate.
Solving blank 2 first (negative) and then blank 1 (positive) reaches the same answer with fewer combinations tested.
ETS gives no credit for partially correct answers. Both blanks must be right to earn the point, so a question where you nail blank 1 and miss blank 2 scores zero. Spend the extra seconds to read both choices together in the sentence before you move on.
Three-Blank Text Completion: Avoiding Cascade Errors
A cascade error happens when a wrong choice for one blank leads you to a choice for a later blank that is consistent with your mistake but wrong for the sentence. The later choice looks right given the early error, so the mistake propagates.
There is no partial credit on three-blank questions either. ETS's rule covers every text completion item: two blanks right and one wrong scores the same as all three wrong. Some guides claim otherwise; ETS's published rule says no. Lock in the blank you are most confident about first, use it as an anchor for the others, and verify all three together before moving on.
Three-blank passages are longer and usually contain a mini-argument: a setup, a development and a resolution or complication. Each blank tends to occupy a different role in that argument, and identifying those roles shows which blank is most constrained.
A full example. "The committee's report, far from being the (i) _______ document observers had anticipated, turned out to be a (ii) _______ critique that (iii) _______ the agency's entire approach to environmental regulation, prompting the agency's director to resign within a week." Blank (i): routine, incendiary, exhaustive. Blank (ii): scathing, tepid, perfunctory. Blank (iii): endorsed, impugned, summarized.
- Start with blank (ii), the most constrained. A critique that led the director to resign within a week was severe. 'Tepid' and 'perfunctory' are mild, so 'scathing' it is.
- Blank (iii) must be what a scathing critique did to the agency's approach. 'Endorsed' is the opposite and 'summarized' is neutral; 'impugned' (called into question) fits.
- Blank (i) must contrast with the scathing critique, because of 'far from being'. 'Incendiary' is in the same direction as 'scathing', so it fails the contrast. 'Exhaustive' is about length, not tone, so it does not set up the contrast either. 'Routine' does.
- Read it whole: far from being the routine document observers had anticipated, the report turned out to be a scathing critique that impugned the agency's entire approach. Answer: routine, scathing, impugned.
The cascade trap here: a test taker who starts left to right and picks 'incendiary' for blank (i) has to make the report the opposite of incendiary, so drifts to 'tepid' and then 'endorsed'. Each later choice is consistent with the first mistake, and the whole question is lost.
Sentence Equivalence: Why It's a Different Beast
Sentence equivalence (SE) looks like a simpler version of text completion: one blank, six choices, pick two. But SE has its own requirement: the two words you choose must each produce a complete sentence, and the two sentences must mean the same thing. Not slightly similar. Not in the same general category. The same.
That requirement is the source of three traps that catch strong test takers:
| Trap | What it is | Example |
|---|---|---|
| Near-synonym | Two words close in meaning but different enough in connotation or register to produce sentences with different nuances. | 'Frugal' and 'miserly' are near-synonyms, but 'frugal' is neutral or approving while 'miserly' is a judgment. In a sentence praising someone's financial habits, 'frugal' fits and 'miserly' does not. |
| Same meaning, different register | Two words that mean essentially the same thing but belong to different levels of formality or different contexts. | 'Deceased' and 'dead' mean the same thing, but a legal sentence takes 'deceased' while a casual one takes 'dead'. If the register constrains one and not the other, they do not produce equivalent sentences. |
| Three plausible words | The sentence admits three words that seem individually correct, but only two produce sentences with equivalent meanings. | You must find the two that produce genuinely equivalent sentences, not the three that individually fit. |
Example: "The athlete's recovery was so _______ that her coaches began training her for competition within six weeks of the injury." The blank needs a word about speed. Candidates: rapid, remarkable, swift, miraculous, dramatic, complete.
'Rapid' and 'swift' produce equivalent sentences: both mean fast, which is what 'within six weeks' requires. 'Remarkable' and 'miraculous' are plausible, but 'miraculous' implies near-impossibility in a way 'remarkable' does not, and neither is specifically about speed. 'Dramatic' describes the nature of the recovery, and 'complete' describes its extent. The answer is rapid and swift.
Always test SE answers together: put word A in and note what the sentence means, then word B, and confirm the two meanings match. If you can state a difference in meaning, one of your choices is wrong. The full method, with six worked examples, is in our sentence equivalence guide.
When You Don't Know a Word in the Choices
You will meet unfamiliar words in the choices. That is expected and manageable, because the sentence's logic narrows the answer before vocabulary comes into play.
Never choose an unfamiliar word just because you don't know it. The tempting thought, 'I don't know this one, so maybe it's the sophisticated answer', takes you out of the logical system and leaves you at the mercy of chance.
Instead, work from the outside in:
- Use your prediction to eliminate the words you do know. If the blank needs a positive, calming word and you recognize three of the five choices as negative, eliminate them. The answer is among the rest, which may include the unfamiliar word.
- Use roots for partial meaning. 'Ameliorate' contains 'melior' (Latin for better), so it means to improve. 'Exacerbate' contains 'acerb' (bitter, harsh), so it means to make worse. Even partial root knowledge often gives you the direction.
- Narrow the field and accept good odds. Eliminate three of the five choices on a one-blank item and you are choosing between two: a 1 in 2 chance instead of 1 in 5.
- For sentence equivalence: if you can identify one correct word, the second must produce the same meaning. Ask which remaining choice, unfamiliar or not, gives the closest equivalent sentence.
Build the vocabulary in context, not as isolated definitions. Start with the 100 most common GRE words, then use the vocabulary trainer, which shows each word in a GRE-style sentence and brings back the ones you miss.
Common Signal Word Reference
These signal words appear often in text completion and sentence equivalence. Knowing the categories by reflex lets you read a blank's direction in seconds.
| Category | What it does to the blank | Signal words |
|---|---|---|
| Continuation | Blank carries the SAME direction as surrounding text | moreover, furthermore, additionally, in fact, indeed, consequently, therefore, thus, as a result, accordingly, because of this, for this reason, in the same way, similarly, likewise, and, also, too, not only...but also |
| Contrast | Blank carries the OPPOSITE direction from surrounding text | however, despite, although, though, even though, yet, but, while, whereas, in contrast, on the other hand, paradoxically, surprisingly, ironically, nevertheless, nonetheless, rather, instead, except, regardless, notwithstanding, in spite of |
| Causal | Blank is the cause or effect of something stated | because, since, so, therefore, thus, consequently, as a result, due to, owing to, leads to, causes, results in, explains, accounts for |
| Emphasis | Blank intensifies or reinforces a quality | especially, particularly, specifically, notably, above all, in particular, most importantly, primarily, essentially, fundamentally, even, clearly, undeniably, remarkably |
| Concession | Acknowledges an opposing point before returning to the main direction; a blank after the concession reverses back to the main direction | admittedly, granted, to be sure, it is true that, while it may be true that, even if, even though one might argue, despite the fact that, although it cannot be denied that |
Concession signals are the trickiest. A sentence that concedes a point and then reverses contains two shifts. Mark both and track which direction applies to each blank. A blank after a concession often carries the opposite direction from the concession itself.
A Three-Week Text Completion Practice Plan
Reading a method does not change your score; drilling it does. A plan that fits around a full-time schedule, about 30 minutes a day:
| Week | Daily work | Goal by the end of the week |
|---|---|---|
| Week 1 | 10 one-blank questions, untimed. For each, write the spine, the signal word and a prediction before looking at the choices. | Your prediction matches the right answer's direction almost every time. |
| Week 2 | 10 mixed questions at the target times (45, 90 and 120 seconds). Log every miss with its cause: direction, intensity, unsupported, sense shift, cascade or unknown word. | You can name the cause of each miss without looking at the explanation. |
| Week 3 | Two timed Verbal sections (12 in 18 minutes, 15 in 23), then a review of every text completion miss against your log. | Sentence items take under a third of each section's time. |
If your log shows mostly 'unknown word' misses, shift time to vocabulary in context. If it shows direction and cascade errors, the fix is the method, not more words. Drill both under real timing in text completion practice, and finish with a free full-length GRE mock to see the method hold up inside a section.