SSC GD Exam Pattern: 80 Questions, 160 Marks, Normalisation Tools

ProSyllabus is an independent study site. We are not affiliated with, or endorsed by, the Staff Selection Commission, the Ministry of Home Affairs or any Central Armed Police Force. Every rule below is quoted or summarised from their published documents, and every sum, ratio or estimate is labelled as our own arithmetic. We print no candidate names or roll numbers.

Four details are easy to miss. The exam is conducted in English, Hindi and thirteen regional languages: Assamese, Bengali, Gujarati, Kannada, Konkani, Malayalam, Manipuri, Marathi, Odia, Punjabi, Tamil, Telugu and Urdu (para 12.1.1). Part D tests English or Hindi, not the regional language. Candidates may not use calculators (para 12.1.10). And the notice prints a single duration of 60 minutes for the paper; it mentions no sectional timer and no sectional cut-off, which is our reading of its silence, so you can spend your 60 minutes across the parts in any order the test interface allows. SSC provides a mock test in the Candidate's Corner of its website (para 12.1.9).

Each right answer earns 2 marks; each wrong answer loses 0.25 marks; an unanswered question scores zero (the notice states the negative marking, para 12.1.2). The penalty is small: it is one-eighth of the marks for a right answer, so the break-even chance of being right is 0.25 divided by 2.25, which is 11.1 per cent: the chance of a right answer above which guessing has a positive expected value (our arithmetic; it is not an accuracy you need overall). The notice does not say how many options a question has; four is the usual layout, and if so a blind guess has an expected value of 0.25 times 2 minus 0.75 times 0.25, which is +0.3125 marks, positive. The table shows the expected marks for one answered question when you can rule out options, on the assumption of four options.

A candidate must score the minimum percentage of the 160 marks to be considered for the physical tests: 30 per cent for UR (48), 25 per cent for OBC and EWS (40), and 20 per cent for all other categories (32). The calculation is on the normalised score "without adding bonus marks to NCC Certificate holders" (para 14.2). NCC certificate holders get incentive marks added to their normalised score for merit: C certificate 5 per cent of the maximum marks (8), B certificate 3 per cent (4.8), A certificate 2 per cent (3.2). The bonus is available only if you claimed it in the application form, is awarded provisionally and withdrawn if the original certificate cannot be produced at document verification, needs a certificate issued on or before 1 January 2026, and is not available to ex-servicemen (para 7.5).

Ties in the final score (including NCC bonus) are resolved in order, per para 15 of the 2026 notice: marks in Part A, then Part B, then Part C, then date of birth with the older candidate higher, then alphabetical order of names. The 2025 write-ups list only Part A, Part B, date of birth and alphabetical order, with no Part C step. We print both and cannot tell from the documents whether the 2026 notice or the 2025 practice will be applied to the 2026 result.

The exam is held over many days and shifts, so papers differ in difficulty and raw marks cannot be compared directly. SSC therefore normalises. The explanatory note of 10 September 2025 says the earlier method, from a notice of 7 February 2019, used the top scores, average scores and the spread of scores of the shifts to compute an adjusted score. SSC then "adopted a new procedure of normalization which is based on equipercentile method", published on 2 June 2025. The 2025 constable exam, notified before June 2025, was normalised by the 2019 formula, which its write-ups cite; the 2026 notice cites the 2 June 2025 procedure (para 12.1.5). That matters for anyone using last year's cut-offs: the numbers are on a different scale of normalisation.

The equipercentile idea is easy to state. A candidate who beat 80 per cent of the people in one shift is matched with a candidate who beat 80 per cent in another shift, and the raw marks of those two are treated as equivalent. The procedure has three steps. Step 1: within each shift, sort the candidates and compute each candidate's percentile P = m/N, where m is the number of candidates with raw score at or below theirs and N the number who appeared, to eight decimal places. Step 2: collect the percentiles of all shifts into one table, sorted, and for each percentile record the raw score in each shift, filling blanks by linear interpolation between the nearest known percentiles. Step 3: the normalised score of a candidate is the average of the raw scores at that percentile across all shifts.

The demonstration data are made up and the tool is a faithful toy, not a prediction of anything: with the two example shifts, a candidate who scored 88 in shift 1 and one who scored 70 in shift 2 stand at the same percentile (0.5), so both normalise to 79, while a raw 99 in the harder shift 2 normalises to 105.5. What to take from it: under the equipercentile method your own rank within your shift is what carries across, so being among the top in a hard shift is worth the same as being among the top in an easy one. A lucky shift does not change your percentile; it changes how the raw marks translate. SSC says the method ensures candidates who faced problems relating to content in one shift are not unfairly ranked against candidates in other shifts.

SSC puts the candidates' response sheets and the tentative answer keys online after the exam and invites objections. For the 2026 exam the key notice is dated 15 June 2026, the challenge window ran from 6 pm on 15 June to 6 pm on 20 June 2026, and each challenged question cost Rs 50 (para 12.1.6 of the notice and the 15 June notice). Objections can be raised only in the language you attempted, though you can view the questions in Hindi and English too. In the challenge module the order of questions and options is the same for everyone in a shift, which may differ from what you saw, but your answers are shown correctly.

SSC's guidelines of 10 April 2026, attached to the constable addendum of 24 April, state what happens after the experts review a challenge. If a question is ambiguous, wrong or has no correct option, it is discarded and its marks are added to every candidate, whether or not they attempted it. If it has more than one correct option, anyone who chose any correct option gets full marks, anyone who chose a wrong option gets the negative mark, and unattempted questions score zero. A question out of the syllabus is discarded and credited to all. A translation or language error is judged separately for that language, so a question can be discarded for Punjabi and stay valid in English. SSC also says only substantive objections count and it will ignore hyper-technical ones.

The Notice (para 14) has five stages: the computer based exam, PET, PST, medical (DME/RME) and document verification. After the exam, male and female candidates are shortlisted state-wise, area-wise and category-wise for the PET/PST, "about 8 (Eight) times the number of vacancies", with more for SSF; after PET/PST, about two times the vacancies are called for DME/DV. On the 25,487 tentative posts that is about 203,900 candidates for the physical test and about 51,000 for medical (our multiplication; the notice says "about"). Final selection and force allocation come from normalised marks with NCC bonus, separately for men and women in every state or UT and area, and force allocation is by merit-cum-preference (para 14.11 and 14.12). The 2025 exam actually shortlisted 394,121 for physical tests (7.3 times its 53,690 posts) and 95,575 for medical (1.8 times).