# Good GRE Score for CS & Data Science MS 2026 | topin

On this page

## Start with the programme, not a combined-score slogan

A combined Verbal-plus-Quant number can be convenient shorthand, but it hides information. Two applicants can both say 325 while presenting very different quantitative evidence. Analytical Writing is also reported separately. Write your scores as Verbal, Quant and Writing before comparing them with any department profile.

First establish [whether the score can be used](/articles/gre-required-ms-us-2026). Stanford MSCS explicitly does not consider GRE scores; Stanford's Statistics/Data Science master's route requires the General GRE, subject to its stated waiver routes. A target that makes sense for one cannot simply be transferred to the other because both are at Stanford.

Next look for programme-level published evidence, with a cohort year and statistic. A mean, median, middle-50% range and minimum requirement answer different questions. An admissions requirement tells you what must be supplied; a class profile describes a particular group. Neither makes an individual admission result predictable.

This guide was checked on 4 October 2026\. It uses a small published historical CMU profile and current ETS percentiles, and labels missing current figures. We do not replace absent data with consultancy averages. The score-setting method below is a planning recommendation, not a department's undisclosed selection rule.

## Published CS and Data Science profiles: keep the year visible

__Official published profile located and checked 4 October 2026; 2019 admissions, CMU comparison PDF dated 15 September 2020; ranges are 25th–75th percentiles among students offered acceptance__
| Programme                                | Cohort / population                 | Quant range | Verbal range | Writing range |
| ---------------------------------------- | ----------------------------------- | ----------- | ------------ | ------------- |
| CMU MS Computer Science                  | 2019 admissions; offered acceptance | 168–170     | 158–165      | 4.0–4.5       |
| CMU Master of Computational Data Science | 2019 admissions; offered acceptance | 168–170     | 157–162      | 3.5–4.5       |
| CMU MS Machine Learning                  | 2019 admissions; offered acceptance | 168–170     | 158–165      | 4.0–4.5       |

These are published figures, but they are not a 2026 entering-class profile. The document is titled Master's Programs Overview and Comparison Data, Winter 2019–2020, with a 15 September 2020 version date. Its footnote describes the score ranges as 25th to 75th percentile and refers to students offered acceptance. That population should not be silently renamed enrolled students.

The PDF includes columns for enrolment and applications beside the score columns. Those nearby numbers do not change the meaning of the score-range footnote. We inspected the visual table because extracted PDF text can place figures beside the wrong programme. The rows above preserve the ranges shown in the document.

A lower endpoint of 168 is not a minimum. A middle-50% interval describes the centre of that historical distribution; it does not say that nobody below the interval received an offer. Equally, reaching the upper endpoint does not establish that an applicant meets every academic or research expectation.

Use these figures as a historical example of score distributions at selective computing programmes. Do not call them current CMU averages or a nationwide target. If a newer departmental profile becomes available, replace the historical row with the new cohort and statistic rather than mixing the two into one synthetic number.

## Where current programme scores are not published

__Current GRE profile data on the official pages checked 4 October 2026; 'not published' is limited to those checked pages, not a claim that no university document exists__
| Programme                                  | Current V / Q / Writing class figures                          | What the checked official page does say                                  |
| ------------------------------------------ | -------------------------------------------------------------- | ------------------------------------------------------------------------ |
| Stanford MS Computer Science               | Not published on checked page                                  | GRE not required and not considered                                      |
| Stanford Statistics/Data Science MS        | No dated section-score class profile published on checked page | GRE required; FAQ gives a vague top-3–5% statement, not a cohort profile |
| Georgia Tech on-campus MS Computer Science | Not published on checked page                                  | GRE optional and considered if provided                                  |
| Illinois MS Computer Science               | Not published on checked page                                  | GRE recommended but not required                                         |
| NYU Courant MS Computer Science            | Not published on checked page                                  | GRE recommended but not required                                         |
| NYU CDS MS Data Science                    | Not published on checked page                                  | GRE optional; GPA information is not a GRE profile                       |
| Duke MIDS                                  | Not published on checked page                                  | GRE optional                                                             |
| Columbia MS Data Science                   | Not published on checked page                                  | GRE optional for 2026 applications                                       |
| UC Berkeley research MS EECS               | Not published on checked page                                  | GRE neither required nor accepted                                        |

Not published is an answer you can plan around. It means a precise programme average cannot responsibly be filled into this table from the checked source. It does not mean zero, low scores, easy admission or no standards. The right next step is to read what evidence the programme asks for and decide whether a score adds anything.

Stanford Statistics' FAQ says average admissions exam scores are in the top 3–5%, but it does not provide a dated cohort, section breakdown or clearly defined class statistic. We therefore do not turn that phrase into a claimed 2026 Quant mean or median. That conversion would create precision the source does not supply.

NYU CDS publishes other admissions information, including academic background and a GPA figure. A GPA profile is not a substitute for a GRE profile. Likewise, a university-wide graduate average or an MBA cohort's GRE average is not a CS or Data Science master's statistic.

If you ask an admissions office for more context, ask whether it publishes a dated profile for the exact degree and whether the figures concern all admits, enrolled students or only score submitters. If no figure is supplied, retain not published. You can choose a personal preparation target without presenting it as an official cutoff.

## What the same GRE number means in Verbal and Quant

__Selected ETS percentile ranks, checked 4 October 2026; reference group tested 1 July 2022–30 June 2025; percent scoring lower__
| Scaled score | Verbal percentile | Quant percentile |
| ------------ | ----------------- | ---------------- |
| 155          | 64                | 37               |
| 158          | 76                | 45               |
| 160          | 82                | 50               |
| 163          | 90                | 60               |
| 165          | 95                | 67               |
| 167          | 97                | 75               |
| 168          | 98                | 80               |
| 169          | 99                | 85               |
| 170          | 99                | 89               |

The table explains why a raw 165 is not the same relative performance in the two measures. In this ETS reference group, it is the 95th percentile in Verbal and the 67th in Quant. Read each measure on its own terms rather than assigning a single combined-score percentile.

A Quant 170 is the maximum scaled score, yet ETS lists its percentile as 89\. Percentiles here count people scoring lower, not everyone scoring lower or equal. A score ceiling with many tied maximum scores can therefore have a percentile below 100\. Do not describe a perfect Quant score as below perfect because its percentile is 89.

The reference group includes GRE test takers, not applicants to your chosen master's programme. An overall percentile does not locate you within a CS admission pool or tell you your admission probability. The historical CMU table and the current ETS percentile table concern different populations and periods.

Use [topin's GRE percentiles tool](/gre/score-percentiles) to inspect other section scores without reproducing the entire conversion table here. Keep the date and population in mind when comparing an older report with a newer table. A percentile can change as the comparison group changes even when the original scaled score stays the same.

## Set a Quant target without inventing a cutoff

Quant deserves its own line in your planning because a combined total can hide an uneven result. That is an interpretation method, not a claim that every CS department assigns the same weight to Quant. CMU MSCS's current admissions page specifically mentions mathematical proficiency when strongly recommending scores; other programmes state their policies differently.

Suppose you have 165 Quant and are considering a programme that accepts the GRE but publishes no current profile. You can accurately say the score is at the 67th percentile in the current ETS table. You cannot accurately say it is below that programme's average without programme data. Decide whether stronger score evidence is worth pursuing by considering the rest of your preparation and the actual deadline.

For a programme with a dated range, compare your Quant score with that range while preserving its age and population. If your goal is to approach the historical CMU 168–170 interval, label that as your chosen target informed by older published data. Do not relabel it the requirement for admission to CMU or every top programme.

Review the mathematical evidence already in your application. Relevant coursework, assessed projects and recommendations may document capabilities that a test score alone cannot. If you are trying to address a specific weakness, explain what a stronger score would demonstrate, then weigh that against improving the supporting academic evidence.

A preparation target should also be testable. Take [a timed diagnostic](/gre/practice-test), identify the recurring reasons for lost marks and compare another fresh test after working on them. If practice results do not improve, extending preparation indefinitely may not be the best use of application time. Keep the degree policy and the submission deadline in the decision.

## Keep Verbal and Analytical Writing in the decision

A technical degree does not justify assuming that Verbal and Writing are irrelevant. Read the profile or requirement for each measure. The historical CMU rows show different Verbal and Writing intervals even where the Quant intervals match, which is another reason not to collapse all programmes into a single total.

Writing is on a separate 0–6 scale in half-point steps. A student saying 330 has not supplied any Writing information. Record it explicitly. When a programme gives a writing requirement or publishes a writing statistic, compare with the right measure rather than an English proficiency score or a Verbal number.

GRE Analytical Writing and English proficiency testing also serve different application requirements. A strong GRE Verbal score is not permission to omit TOEFL or IELTS unless the degree's official instructions say so. Treat those checks separately, including their waiver rules and score-age conditions.

For preparation, use the score breakdown to decide where extra work is likely to matter. If your Quant evidence is already strong and [your timed essay](/gre/analytical-writing/analyze-an-issue) is incomplete, spend some sessions on planning and finishing an argument. If your reading accuracy falls under time pressure, inspect that separately from your vocabulary knowledge.

The purpose is a balanced piece of application evidence, not a cosmetically attractive sum. If a programme says it will not consider any GRE scores, further polishing any measure for that programme has no score-reporting benefit. If it considers them, use the complete result and the exact submission instructions.

## A practical target worksheet for your shortlist

__Illustrative applicant decisions, prepared 4 October 2026; these are planning examples, not published cutoffs or admission predictions__
| Situation                                     | Evidence available                     | Sensible next action                                                            |
| --------------------------------------------- | -------------------------------------- | ------------------------------------------------------------------------------- |
| Target programme excludes GRE                 | Explicit not-considered policy         | Invest effort in required application evidence                                  |
| Score required; no class profile              | Requirement and your diagnostic        | Set separate section targets; plan reporting before the deadline                |
| Optional score; current class profile exists  | Exact cohort, statistic and your V/Q/W | Compare like with like, then decide whether submission supports the application |
| Optional score; only historical profile found | Old distribution and current policy    | Label the profile's year; avoid calling it a current average                    |

Create one row per real application with the programme name, score policy, profile source, cohort year, statistic and your section results. Leave a blank or write not published when a source does not give data. An honest gap is more useful than a precise-looking number that cannot be traced.

Then identify your binding constraint. It may be a required-score application deadline, an attainable section improvement, available study time or [the fee for another attempt](/articles/gre-fees-2026). Work backwards from that constraint. A higher target without time to receive and submit the score is not a usable plan.

If you are comparing optional-score programmes, avoid sending to every recipient automatically. Evaluate whether the complete result adds evidence to each application and whether the department requires official receipt now or allows self-reporting. Those instructions can change both your timing and your reporting cost.

Keep two different columns: published programme statistic and personal target. The first must come from a named official source; the second is your decision. This prevents a reasonable aspiration, such as moving closer to a historical range, from gradually becoming a fictitious university cutoff in your notes.

## When a retake is worth considering

Consider another attempt when a programme will use the score, the desired improvement addresses a real weakness and your preparation has changed enough to make improvement plausible. This is a judgement about your work, not a guaranteed score increase. Repeating the same preparation with a higher hoped-for total gives you little new evidence.

Check the calendar before booking: ETS allows one General Test every 21 days, up to five in any rolling 365 days, and official scores arrive 8–10 days after testing. Our guide to [GRE score validity and retakes](/articles/gre-score-validity-retakes) covers ScoreSelect and reporting deadlines in detail.

If you already satisfy your chosen score target, compare the potential gain with the application work still unfinished. A clearer project account, stronger evidence of fit or a complete prerequisite record may deserve attention. No score from the tables above guarantees admission, and no missing class profile justifies inventing one.

The best stopping point is a documented decision: which programmes can use the result, what you know from published data and why you will submit, retake or stop. That is more reassuring than chasing a single number that changes from one unsourced article to the next.

## Try a full GRE mock free

Timed like test day, every section scored, every answer explained.

[Take the free mock ](/gre/login)

## FAQs

Is 320 a good GRE score for MS in Computer Science?

The total alone is insufficient. Check the Verbal, Quant and Writing breakdown, the programme's GRE policy and any dated published class profile.

Do I need 330 for a top CS master's?

There is no verified universal 330 requirement. Some selective CS programmes do not consider the GRE at all.

Is 165 Quant good for Data Science?

It is the 67th percentile in the ETS reference group checked here. Whether it supports your application depends on the exact programme and its published information.

Why is a 170 GRE Quant score only the 89th percentile?

ETS's percentile counts test takers scoring lower, excluding ties. The maximum scaled score can therefore have a percentile below 100.

What is Stanford's average GRE score for MSCS?

No current GRE class figures are published on the checked MSCS FAQ, and the programme does not consider GRE scores. We do not supply an invented average.

What GRE scores does CMU publish for MSCS?

An official historical 2019 admissions table gives middle-50% ranges of Quant 168–170, Verbal 158–165 and Writing 4.0–4.5\. These are not a current class profile or minimums.

Does not published mean there is no GRE cutoff?

It means the checked page provides no dated class-score figures. A requirement or cutoff must be checked separately; absence of a profile does not prove absence of a rule.

Should I retake if my GRE is below a published average?

An average is not a minimum. Consider the programme's policy, your complete application, realistic improvement and whether a new result can arrive before the deadline.

## Sources (checked 4 October 2026)

* [CMU: official 2019 master's admissions comparison profile, version 15 September 2020](https://www.cs.cmu.edu/education/resources/masters-overview-comp-data-2020-09-15.pdf)
* [ETS: current General Test Interpretive Data, July 2022–June 2025 reference group](https://www.ets.org/content/dam/ets-india/pdfs/gre/gre-guide-table-1a.pdf)
* [ETS: Guide to the Use of Scores, separate measures and holistic interpretation](https://www.ets.org/content/dam/ets-org/pdfs/gre/gre-guide-to-the-use-of-scores.pdf)
* [Stanford CS: MS FAQ, no GRE consideration and no class figures on checked page](https://www.cs.stanford.edu/admissions/masters-admissions-frequently-asked-questions)
* [Stanford Statistics: MS FAQ and unqualified average-exam statement](https://statistics.stanford.edu/msadmissions/ms-admissions-faq)
* [Georgia Tech: MSCS requirements, no class-score figures on checked page](https://www.cc.gatech.edu/ms-computer-science-admission-requirements)
* [Illinois Siebel School: application requirements, no class-score figures on checked page](https://siebelschool.illinois.edu/admissions/graduate/applications-process-requirements)
* [NYU Courant CS: MS FAQ, no class-score figures on checked page](https://cs.nyu.edu/dynamic/masters/prospective-overview/masters-program-admissions-faqs/)
* [NYU CDS: MS requirements, no class GRE profile on checked page](https://cds.nyu.edu/admissions/masters-in-data-science-admissions-req/admissions-requirements/)
* [Duke MIDS: requirements, no class-score figures on checked page](https://datascience.duke.edu/admissions/requirements/)
* [Columbia DSI: MS page, 2026 optional policy and no class GRE figures](https://datascience.columbia.edu/education/ms-data-science)
* [Berkeley EECS: research admissions, no GRE consideration](https://eecs.berkeley.edu/academics/graduate/research-programs/admissions/)
* [CMU CSD: current MSCS recommendations concerning mathematical proficiency](https://www.csd.cmu.edu/academics/masters/admissions)
* [ETS: score scales and availability](https://www.ets.org/gre/score-users/about/general-test/scoring.html)
* [ETS: General Test retake rules](https://www.ets.org/gre/test-takers/general-test/register.html)

## Related articles

[All GRE study guides ](/gre)

* [GREAdmissionsIs the GRE required for MS?Check 30 US master's programmes in CS, data science, electrical and mechanical engineering: GRE requirements, waivers, intake caveats and department sources.Updated 4 Oct 2026](/articles/gre-required-ms-us-2026)
* [GREFormat & test dayGRE at home checklistPrepare for GRE at home in 2026: advance identity verification, two cameras, equipment and room rules, plus a practical 72-hour to check-in checklist.Updated 4 Oct 2026](/articles/gre-at-home-2026)
* [GREFees & retakesGRE fees and savingsThe GRE costs $249 from August 2026\. See every extra charge, the 4% service fee, what we know about the INR price and how the $100 Fee Reduction Program works.Updated 4 Oct 2026](/articles/gre-fees-2026)
* [GREFees & retakesValidity and retakesGRE scores last five years. Learn the 21-day retake rule, five-test rolling limit, ScoreSelect choices, reporting deadlines and cancelled-score options.Updated 4 Oct 2026](/articles/gre-score-validity-retakes)