Loading...
Loading...
Straight answer: aim for 90 to 95 of the 100 questions in GS Paper 1, not all 100. That covers everything you know plus the questions where you can eliminate an option or two and lean one way. The last handful are usually pure blind guesses, and with one-third negative marking those cost you more than they pay. Your real number is set by your accuracy, not by how brave you feel on the day.
There is no single magic number, but there is a sane band: attempt 90 to 95, leave the 5 to 10 where you are guessing blind. Every wrong answer in GS Paper 1 costs you 0.66 marks, every correct one earns 2, and a blank costs nothing. So the question is never “how many bubbles can I fill,” it is “how many can I fill with at least a fighting chance.” Get your accuracy up and the safe attempt count rises with it. That is the whole page in three lines.
GS Paper 1 is 100 questions, 2 marks each, 200 marks in two hours. The instinct of a nervous aspirant is to fill all 100, on the logic that more attempts must mean more marks. It does not. The last few you force in are the ones you have no basis for, and under negative marking a random tick is a slow leak, not a lottery ticket. The mentor is blunt about the ceiling when a student told him he planned to attempt everything.
Avoid. 90-95 is the golden range imo!
Be decisive, quick and move with immense self belief. Mark then and there in the OMR and aim to finish atleast one read within 1hr45mins having marked around 80 questions in OMR by then. Don't attempt 100, stick to 90-95! Be your own Buddha during those 2 hrs, No Mara shall trouble you!
Ninety to ninety-five is not a rule carved in stone, it is a band. A strong, quick reader who can clear an old paper cold can sit at the top of it. If the paper feels heavy or your accuracy has been wobbly in practice, drop to the lower end and protect your net. Use the table below to place yourself honestly.
| Attempt band | Who it fits | Why |
|---|---|---|
| 95 to 100 | Almost nobody | Forces in pure blind guesses that bleed net marks under negative marking |
| 90 to 95 | Most well-prepared aspirants | Covers everything you know plus fair educated guesses; the golden range |
| 80 to 90 | Lower accuracy, or a hard paper on the day | Protects your net score when a lot of the paper feels shaky |
| Below 80 | Rarely the right call | Usually leaves easy educated-guess marks on the table |
You cannot decide how many to attempt without knowing what a wrong answer costs. UPSC deducts one-third of a question’s marks for a wrong answer. On a 2-mark question that is 0.66 marks gone. A correct answer earns the full 2, and a blank earns and costs nothing. So a wrong answer does not just fail to add 2, it swings you 2.66 marks below the version of you who got it right. That gap is the whole reason a careless guess hurts.
Now put a probability on it. Filling an option is a bet, and the payout depends on how many options you have honestly ruled out. Here is the expected value of a single guess in each situation, using the real 0.66 penalty.
| Your situation | Odds of the right answer | Expected marks per guess |
|---|---|---|
| Blind guess, all 4 options open | 1 in 4 | about 0 (roughly break-even) |
| One option eliminated | 1 in 3 | about +0.23 marks |
| Two eliminated, a genuine 50-50 | 1 in 2 | about +0.67 marks |
Read that table twice. A blind guess is roughly a coin toss with the house, close to zero over many questions and not worth the anxiety. The moment you knock out one option, the bet turns clearly positive, and a real 50-50 pays well. The founder does not agonise over a single hard question, he treats the small penalty as a cost of doing business.
You will always clear the cutoff irrespective. Pick your battles wisely. I'll take a neg dot six six any day!
“Neg dot six six” is his shorthand for that 0.66 deduction. His point is that you can lose a few of these small bets and still clear the cutoff comfortably, so long as you are winning the ones where you had an edge.
This is where most of your 90 to 95 attempts come from. You will know maybe 60 to 75 questions with real confidence. The gap between that and your final count is filled by educated guesses, and the skill is telling an educated guess apart from a blind one. If you have ruled out one or two options and you feel a pull towards a particular answer, that pull is usually worth backing.
Attempt all of these in the first round unless it’s a pure 50-50! 90% of times, we lean towards one.
Note the discipline in that: attempt it in the first round, do not park it for a second visit that eats your time. The one exception he draws is the pure 50-50 where you have genuinely no lean. Most of the time you do lean, and that lean, built from months of decoding past papers, is the edge that makes the guess pay. Building that instinct is the entire idea behind the Sherlocking method: basics in, application on top, and elimination drilled until it becomes reflex rather than a hunch you second-guess.
One warning the mentor repeats: an elimination trick only counts if you could have used it in the hall without already knowing the answer. A clever line that only makes sense once the key is out is not a skill, it is hindsight. Real educated guessing is the reasoning you can run cold, under pressure, on a question you have never seen.
Leaving questions is a skill, not a failure. The 5 to 10 you skip are what let you attempt the other 90 to 95 with a clear head. Two situations call for a skip. The first is when your own elimination clues fight each other and point to different options.
Very difficult to get correct. I would advise you to leave such questions, because a lot of conflicting heuristics.
When two lines of reasoning each look sound but land on different answers, you are not close to the answer, you are being pulled in circles. That is a leave, not a fight. The second situation is the handful of genuinely ambiguous questions UPSC sets every single year.
I don't agree with UPSC's key either. That's why I say, there will always be 2-3 questions every year with ambiguous answers. Stick to basics and don't waste your energy in time/emotional sinks like these. Iske alawa bhi you would have enough scope to clear the cutoff comfortably.
Two or three questions a year will have a debatable key even the toppers argue over. Sinking five minutes and a chunk of your calm into one of them is a bad trade. There is always enough scope in the rest of the paper to clear the cutoff. Use this checklist to make the attempt-or-leave call fast.
| Signal in the question | Decision |
|---|---|
| You can eliminate two or more options | Attempt, mark it in the first round |
| You can eliminate one and you lean one way | Attempt as an educated guess |
| A genuine 50-50 with no lean at all | Usually leave it |
| Two heuristics point to different answers | Leave it, the logic is fighting itself |
| One of the year’s designed ambiguous questions | Leave it, protect your net |
Two aspirants both attempt 90 questions. One clears with room to spare, the other misses. The difference is not the attempt count, it is accuracy, the share of attempts they get right. Hold attempts fixed at 90 and watch how the net score moves with accuracy alone.
| Attempted | Accuracy | Correct | Wrong | Net (of 200) |
|---|---|---|---|---|
| 90 | 80% | 72 | 18 | about 132 |
| 90 | 70% | 63 | 27 | about 108 |
| 90 | 60% | 54 | 36 | about 84 |
| 90 | 50% | 45 | 45 | about 60 |
Same 90 attempts, and the net score nearly doubles from the bottom row to the top. Cutoffs shift year to year, so chase the accuracy, not a target number. And accuracy comes from thinking plainly, not from cleverness.
The simpler you think, the higher the likelihood of arriving at the right answer. Such simple papers made difficult by anxieties projected on us by the market.
He learned this the hard way, and it cost him a full year before it clicked. Overthinking a question from your own supposed expertise is one of the fastest ways to talk yourself out of the right answer.
Always like a layman. Everytime I have used my "science background", I have taken a negative. My hit rate in S&T in 2020 was so low because of that, which costed me an year.
Now I think like a 16yr old non science background student and viola, my hit rate has improved drastically!
Where does the accuracy itself get built? Not in the hall, but in how you work your previous-year questions beforehand, taking each one apart until you know why every wrong option is wrong. If you are unsure how far back to go, our guide on how many years of PYQ to solve lays out a tiered plan. That decoded practice is what turns a nervous 60% attempt into a calm 75% one.
Knowing your target attempt count is half the job. Executing it under a ticking clock is the other half. The mentor keeps the mechanics simple: go in order, mark the OMR as you go, and finish one full read of the paper by the one-hour-forty-five mark with around 80 questions already bubbled in. That leaves a clean window to revisit the ones you flagged. On the question of order, he is clear.
Contextual
Sequential is best imo
Pick and choose will lead to pilferage of time. Flow ensures optimisation of the most imp resource i.e. Time!
Jumping to your strong subjects first feels smart but costs you time in flipping pages and hunting, and time is the scarcest thing you have in those two hours. A steady sequential sweep, deciding attempt-or-leave on each question as it comes, is what makes the 90 to 95 target reachable without a scramble at the end. That judgement is a muscle you build under exam conditions, which is exactly what a sensible number of mock tests is for.
Aim to attempt roughly 90 to 95 of the 100 questions in UPSC Prelims GS Paper 1, not all 100. That range covers everything you know outright plus the questions where you can eliminate one or two options and lean one way. The last 5 to 10 tend to be pure blind guesses, and with one-third negative marking those quietly bleed your net score. The exact number is personal: a strong, fast reader can push to 95, while a shakier attempt is safer around 85 to 90.
Educated guessing is worth it; blind guessing is not. If you can eliminate even one option and you lean towards an answer, the odds tilt in your favour and the expected marks per guess turn positive. A pure blind guess across four open options is roughly break-even after the 0.66 penalty, so it is not worth the risk to your score or your headspace. The rule is simple: guess when you have a reason, leave when you do not.
A safe number is about 85 to 95 attempts for most aspirants. Below 80 you usually leave easy educated-guess marks on the table; above 95 you start including pure blind guesses that drag your net down. Where you land inside that band depends on your accuracy and how the paper feels on the day. The UnlockIAS mentor’s own golden range is 90 to 95.
UPSC Prelims deducts one-third of a question’s marks for every wrong answer. Each GS Paper 1 question carries 2 marks, so a wrong answer costs 0.66 marks, while a correct one earns 2 and a blank costs nothing. Across 100 questions worth 200 marks, a wrong answer swings you 2.66 marks away from where a correct answer would have put you, which is exactly why reckless guessing is expensive.
No, attempting all 100 is usually a mistake. Forcing the last handful means answering questions where you have no real basis to choose, and one-third negative marking punishes precisely that. Most well-prepared aspirants score higher by attempting 90 to 95 and leaving the genuine blind spots. The goal is the highest net score, not the most bubbles filled.
It depends on how many you get right, because only your net score counts. As a feel for the math, at 90 attempts and 70% accuracy you get 63 right and 27 wrong, landing near 108 out of 200. Every wrong answer trims 0.66, so 15 loose guesses cost you about 10 marks. You do not need a fixed wrong-answer budget; you need enough correct answers and fair educated guesses to clear the year’s cutoff comfortably.
Knowing when to guess and when to leave a question is built through timed practice, not luck. The Sherlocking test series trains that judgement under real exam conditions.
Sources: UPSC Civil Services (Preliminary) GS Paper 1 has 100 questions of 2 marks each (200 marks, two hours), with one-third negative marking, which is a deduction of 0.66 marks for each wrong answer and no deduction for a blank. Confirm the current pattern against the latest official notification at upsc.gov.in . The 90-to-95 attempt range and the guess-or-leave calls reflect the UnlockIAS mentor’s method and experience, not an official rule; the expected-value and net-score figures are arithmetic on the published marking scheme. Mentor quotes are reproduced verbatim from the UnlockIAS community archive.
Last updated: July 2026.