Civil Service Judgement Test Pass Mark and Scoring Explained

Before the scoring makes sense you need the shape of the test. Everything in this table is from the GOV.UK guidance page for the Civil Service Judgement Test, except the row on the answer options for part 1, which comes from the official practice test.

The official name is the Civil Service Judgement Test. GOV.UK describes it as “an online situational judgement test”, which is why candidates also search for it as the Civil Service situational judgement test. It is one of the commonly used Civil Service online tests: the GOV.UK online-tests page (updated 27 January 2026) lists 8, while the Civil Service Careers page on assessments and interviews still says “There are 7 commonly used tests”. Both list the Civil Service Judgement Test. The job advert says whether a vacancy uses it.

Two things follow from those sentences. First, the result you see is relative. It places you against other people, not against a fixed number of right answers. Second, difficulty is part of the calculation, so two candidates who see different scenarios (the page says you only see scenarios relevant to the job level you applied for) are not simply being counted on answers matched.

What the pages do not explain is the mechanics underneath: how each of your ratings is compared with the intended rating, whether a rating one step away from the intended one earns partial credit, or how difficulty is weighted. None of that is published. Any website that tells you a near miss is worth half a mark, or that a wrong rating loses marks, is describing something the Civil Service has not put on an official page we could find.

The test has two parts, and the first is easy to overlook. In the words of the guidance: “In the first part, you will be asked about your typical behaviour and preferences at work. You must rate how ambitious, motivated and flexible you are in the workplace.” It then states the weighting: “This section is self-assessment, and makes up 15% of your overall CSJT score.”

Our arithmetic: if the self-assessment is 15% of the overall score, the rest of the test accounts for the other 85%. The page does not print the figure 85, and it does not say how that share is divided between behaviours or scenarios. It does add a warning that matters more than the weighting: “You should answer the questions honestly, as you may be asked about them at interview.”

In the second part you read or watch a workplace scenario and give an effectiveness rating to each action. These are the four options and their official descriptions, word for word.

The guidance is explicit that the ratings are not a ranking exercise: “When reviewing an action, look at it independently from the others. Do not assume that only one action is either effective or ineffective - you can rate more than one action the same.” So a scenario can have two actions that deserve the same rating, and you are not required to use each rating once.

This is the part most pages blur. The official text describes two separate hurdles, decided at two different times.

Read together, those lines mean a candidate can be told they met the minimum standard on the day they finish and still be told, after the deadline, that they did not reach the pass mark set for that particular job. That is not a contradiction. The first message is about the level of the job; the second is about one vacancy and the field of applicants it attracted.

The online-tests page describes the first hurdle as a percentile: “Meeting the minimum required percentile is no guarantee of an invitation to continue the selection process.” So the minimum standard is a percentile figure, but the page does not say what it is for any grade.

The guidance names two things: “the impact on protected groups, and the number of invitations to the next stage”. Our reading of that sentence: a vacancy with many applicants and few interview places can end up with a higher pass mark than the same grade of job with fewer applicants, because the number of people the recruiter can take forward is one of the two stated inputs. That is our inference from the wording, not a rule the page spells out, and the page gives no example figures.

The page does not give a total number of scenarios. It gives a rule: “There are three scenarios for each behaviour, and each scenario asks you to rate four actions.” It also says the test measures “specific Civil Service Behaviours, which have been selected as important for success in your chosen job”, without saying how many behaviours a test covers. On timing it says: “The test is not timed, however most people take between two and four minutes to answer one scenario,” and later, “While the test is not timed, we recommend allowing at least an hour to complete it.”

The same arithmetic as a table, for readers on a page where the calculator does not load. The first column is the number of behaviours; there are nine Civil Service Behaviours in total, but nine is not a stated test length.

This table is our multiplication of the two official figures. It leaves out the self-assessment section and the practice questions, which is one reason the official advice of at least an hour is the safer plan.

For a candidate below the minimum standard the guidance has one line: “If you do not pass at the minimum required standard for a job level, you can retake the test during any future applications.” There is no waiting period on the page, and no limit on the number of future applications.

For a candidate who passes, note one further check. The page says: “Your test answers should be your own; you must not ask for input from anyone. Hiring managers may require candidates invited to interview to retake the test under supervised conditions. You will learn whether you need to do this if you are invited to interview.” A score obtained with help is therefore a score you may be asked to repeat in front of someone.