Why "Safe" Needs a Definition
A safe score is not the cut-off. The cut-off is known only after the final result, ten months after the exam, and it moves every year with the paper and the vacancies. A safe score is a target set before the exam that clears the cut-off in every plausible year, including a hard one. Defining it needs three numbers: where the cut-off usually sits, how far it can move in one year, and the margin that covers the movement plus a bad day.
Number 1: The Band
| Period | General cut-off, GS Paper I | Band |
|---|---|---|
| 2016 to 2017 | 116.00, 105.34 | Above 105; an older, easier regime |
| 2018 to 2020 | 98.00, 98.00, 92.51 | 92 to 98 |
| 2021 to 2025 | 87.54, 88.22, 75.41, 87.98, 92.66 | 85 to 93, with 2023 as the outlier |
The current band, read from the last five cycles and excluding the 2023 outlier, is 85 to 93. The band has drifted down over the decade because the paper's style has shifted towards multi-statement questions and current-static pairings that lower scores across the board; the bar did not fall because the competition eased.
Number 2: The One-Year Swing
The largest single-year moves in the decade were 2016 to 2017, down 10.66, and 2022 to 2023, down 12.81, followed by 2023 to 2024, up 12.57. A swing of about 13 marks in either direction is therefore the realistic worst case. A target that sits only five marks above the band is inside the swing; a target thirteen or more above it is outside.
Number 3: The Margin
Two margins are needed: one for the paper and one for you. The paper margin is the swing, about 10 to 13 marks. The personal margin covers the gap between your mock average and your exam-day score, which for most candidates is a loss of 5 to 10 marks from nerves, misreads and a changed answer or two. Together that is 15 to 20 marks above the band.
| Target | Clears the cut-off in | Verdict |
|---|---|---|
| 90 | 2021, 2022, 2023, 2024 only; not 2025, not 2016 to 2020 | Inside the band; not safe |
| 100 | Every year from 2018 to 2025 | Safe for the current regime; the working target |
| 110 | Every year except 2016 | Comfortable; the target for candidates with a strong base |
| 120 | Every year of the decade | Top-rank territory; the target for toppers, not a safety requirement |
The working answer: 100 in GS Paper I, with a plan to reach 110. Reserved-category candidates should take their own band from the category-wise table and add the same 15 to 20 marks rather than planning for the category cut-off itself, which rose sharply in 2024 and 2025.
Translating 100 into Correct Answers
GS Paper I has 100 questions of 2 marks each with a penalty of one-third of the marks, 0.67, for a wrong answer. The score depends on both the number attempted and the accuracy.
| Attempted | Correct | Wrong | Score |
|---|---|---|---|
| 60 | 52 | 8 | 98.7 |
| 70 | 58 | 12 | 108.0 |
| 75 | 60 | 15 | 110.0 |
| 80 | 62 | 18 | 112.0 |
| 85 | 62 | 23 | 108.7 |
| 90 | 62 | 28 | 105.3 |
Three things the table shows. First, 100 needs about 52 correct answers at 85 per cent accuracy on 60 attempts, or 58 correct at 83 per cent on 70. Second, beyond about 80 attempts, extra attempts at the accuracy most candidates can sustain reduce the score. Third, the same 62 correct answers score 112 with 18 wrong and 105 with 28 wrong; accuracy on the marginal question decides the margin, not the number of questions known.
Setting Your Attempt Count
- From the last five mocks or real papers, record attempts, correct and wrong for each
- Compute accuracy on attempted questions; if it is above 80 per cent, you are under-attempting and can add questions; if below 65, you are over-attempting and losing marks to negatives
- Set the attempt count where your marginal accuracy, the accuracy on the questions you are least sure of, is still above 50 per cent, since a question answered at 50 per cent accuracy is worth 0.67 marks and at 40 per cent is worth 0.27
- Hold the count for three mocks; adjust only if accuracy on attempted questions changes
- For most candidates the count lands between 70 and 80
Checking Yourself Against the Target
Mock series are often calibrated harder than UPSC, so a mock score of 90 may correspond to a real-paper score of 100 or more. The only trustworthy check is the actual previous year paper under exam conditions: the 2024 paper against 87.98 and the 2025 paper against 92.66. A candidate scoring 100 or more on both, three months before the exam, is on target; one scoring 85 is inside the band and needs the diagnosis of why.
What Moves the Target During the Year
- If the year's notification shows a sharp fall in vacancies, add five marks; fewer posts raise every cut-off
- If the paper is widely reported as very hard on the day, the cut-off will fall and a 90 may clear; but this cannot be planned for
- If your category's cut-off rose last year, as SC and ST did in 2024 and 2025, plan for the new level, not the old
Common Mistakes
- Treating the latest cut-off as the target; it is a floor from a paper you will not sit
- Planning for the 2023 outlier as the new normal
- Attempting 90 or more questions at 65 per cent accuracy to "reach 100"; the negatives make it 105 at best and 90 on a bad day
- Reading mock scores as real scores without checking calibration
- Adding CSAT marks to the target; only GS Paper I counts
Know Your Number on IAS Ladder
The attempt-count method depends on knowing your accuracy on the questions you are least sure of, which IAS Ladder records through confidence tagging on every previous year question and chapter-wise MCQ you solve, alongside accuracy by topic. Year-wise UPSC Prelims papers on the app give the real-paper check against the published cut-off. In English and Hindi. Download IAS Ladder and set your target from your own data.
Practicing UPSC Prelims PYQs?
Practice 45+ years of topic-wise previous year questions (1979–2026) with in-depth explanations on our mobile app.
Never miss a daily update
Join our official Telegram channel or WhatsApp group for daily current affairs and test alerts.
Frequently Asked Questions
Tags
Keep reading
More in UPSC CSE
- Prelims How to Approach UPSC Prelims Mock Tests: When to Start, How Many, and How to Analyse
- Prelims UPSC Prelims Revision Strategy for the Last 30 Days: A Day-by-Day Plan
- Prelims How to Prepare for UPSC Prelims: Subject-Wise Strategy for GS Paper I
- Prelims How to Solve UPSC Previous Year Questions Effectively
- Prelims How to Improve Accuracy in UPSC Prelims: Beat Negative Marking
- Optional & Subjects Colonial Land Revenue: Permanent, Ryotwari and Mahalwari