Loading...
Loading...
Keep it simple: the last 10 years are your non-negotiable core, push to 15 if the calendar allows, and hold CSAT to the last 5 to 7. But the year count is the easy half of the question. What actually moves your score is how you take each paper apart, one option at a time, which is where most aspirants quietly lose marks.
Solve at least the last 10 years of Prelims PYQs properly, and stretch to 15 if you have the runway. Extra years help only when you are analysing each paper, not just checking answers. Past roughly 15 years the returns thin out for GS, because the syllabus and question style have drifted. For CSAT, the last 5 to 7 years is enough. If you take one line from this page, take this: the number of years matters far less than the depth you go into each one.
Start with the last 10 years and treat that as the floor, not the ceiling. Ten years is enough to see which themes UPSC keeps circling back to and how it likes to frame a tricky statement. If you have six months or more of clear runway, extend to 12 to 15 years, because the extra papers sharpen your sense of how deep a topic gets asked. Only base-builders who are starting cold with a full year should go back 20 to 25 years, and even then the older papers are there to teach elimination and static fundamentals, not current-affairs weighting.
Why not solve every paper ever set? Because a 2005 GS paper was a different animal: heavier on flat static facts, lighter on the layered, statement-based questions that dominate now. You mine old papers for the durable material and read the recent decade far more closely. The founder is blunt about why PYQs beat chasing the whole syllabus in the first place.
"UPSC's syllabus is a vast ocean, anything can be asked": On paper, yes, but PYQ Analysis tells you the favoured themes.
That is the real reason ten focused years beat twenty skimmed ones. Pull the papers from the UPSC Prelims PYQ bank and sort them by year before you start, so you are never solving blind.
| Scope | Years to cover | Who it is for | What you are mining for |
|---|---|---|---|
| Minimum viable | Last 10 years | Repeat aspirants, tight timelines | Repeated themes, elimination patterns, option traps |
| Recommended | Last 12 to 15 years | Most serious first and second timers | Theme frequency, how deep each topic is asked, statement framing |
| Deep base-build | Last 20 to 25 years | Early starters building from zero | Full pattern map, rare themes that still resurface, static backbone |
| CSAT (Paper 2) | Last 5 to 7 years | Everyone, non-negotiable | Solvable question types, time-cost per set, what to skip |
Short answer: no, and yes, depending on what you mean by “solving.” If solving means ticking options and checking the key, that alone will not clear Prelims. If solving means taking each question apart until you know exactly why the wrong options are wrong, that is most of the battle. The gap between those two habits is the gap between an aspirant who reads a paper and one who decodes it.
Here is a student asking whether the slow, detailed way of working through PYQs is worth the hours. The founder does not hedge.
That's the only thing which has RoI, unlike reading current affairs booklets, or referring multiple sources for obscure subjects.
So the PYQs are the input, and the analysis is the output that actually scores. That decoding habit even has a name inside UnlockIAS.
It's a fancy name we've given to our methodology to clear the exam, which focuses on Basic sources, application of the basic knowledge and exhaustive PYQ analysis all while working on the psychological frame to optimise the performance.
You can read the full idea behind the Sherlocking method, but the one-line version is this: basics in, application on top, PYQs analysed to exhaustion. Solving alone is where most people stop, and it is also where most people plateau.
Once you have picked your window of years, the next fork is whether to solve papers whole (yearwise) or grouped by subject (subjectwise). Both have a place, and the honest choice depends on where you are in your prep.
Early on, while you are still finishing a subject, subjectwise practice tells you whether your coverage of, say, modern history or polity actually holds up. Once your basics are in, the founder prefers full papers, timed, because that is the only way to feel the real thing: a mixed bag where you must judge, on the spot, what to attempt and what to leave.
That's why I prefer doing questions yearwise than subjectwise. Yearwise, you would know the cutoff for each year. And it'll be apparent as to what you can afford to leave. But if something has already featured in PYQs, read that now, in case a question on the same theme appears.
You're doing great. Feeling what everybody feels. Keep at it.
Notice the two gifts of the yearwise method. You see each year’s cutoff, so you learn what a clearing attempt actually looks like, and you build the nerve to leave questions without panic. If you are still assembling your foundation, pair this with a self-study plan for beginners so you are not solving whole papers on an empty base. Use the table below to pick the mode that fits your week.
| Approach | Best when | The catch |
|---|---|---|
| Subjectwise | You are still finishing a subject and want to test coverage | Hides the real cutoff and the mixed judgement a full paper forces |
| Yearwise, timed | Basics are done and the exam is near | Feels brutal at first; you will leave many questions, and that is the training |
This is the step people skip, and it is the whole game. Analysing a previous-year question does not mean reading a ready-made explanation and nodding along. It means rebuilding the reasoning yourself, so the skill transfers to a question you have never seen. UnlockIAS keeps answer-by-answer explanations off these pages on purpose, because a handed-over answer teaches you nothing about the next paper. The founder puts it sharply.
But the real value lies in understanding 'HOW' to arrive at those answers - a skill that remains critical. Often questions come from derived themes from PYQs. So, the exact question coming next year might be tweaked, rendering the crammed answer irrelevant, but the process of 'HOW to create the answer' often remains the same.
Read that twice. UPSC rarely repeats a question word for word. It repeats the theme and tweaks the wording. Memorise a specific answer and you are exposed the moment the framing shifts. Learn the how, and a reshaped question is still yours. Here is a repeatable way to work a single question without ever leaning on a printed answer.
| Step | What you do |
|---|---|
| 1. Attempt cold | Solve the question under time before you look at any answer or key |
| 2. Mark the decision | Note which options you eliminated, why, and where you were guessing |
| 3. Trace the theme | Name the syllabus theme and check how often it has shown up before |
| 4. Fix the leak | Return to your basic source only for the exact gap the question exposed |
| 5. Log the trap | Record the format (assertion-reason, pairs, chronology) so you catch it next time |
Run this loop on every question you get wrong, and a stubborn set of ten papers turns into a personal map of your own leaks. If your last attempt fell short and you are not sure why, this same loop sits at the core of diagnosing a prelims failure.
Aspirants ask for a magic number of years because they want permission to stop reading and start trusting themselves. The better signal is not a year count at all. It is your hit-rate on papers you have already analysed. When you can sit an old paper cold and clear it comfortably, your PYQ work is doing its job.
Focus on overarching plan and completion. Plus have an objective barometer of completion like 80%+ hit rate in PYQs so that you can exit the reading Chakravyuh challenge.
That 80% figure is a barometer for your own readiness, not a claim about how many questions repeat in the exam. And if you are starting late with no PYQ work behind you, do not freeze. There is a sane way in.
You can either enroll in the course or analyse yourself atleast 2015-2021 on the lines of the free 2022 free analysis videos!
Analyse a solid block of recent years properly, and you will be in far better shape than someone who owns every compilation on the shelf and has decoded none of them.
Solve at least the last 10 years of UPSC Prelims previous-year questions as your core, and extend to 15 years if you have six months or more to prepare. Ten years is enough to spot the themes UPSC keeps returning to and how it frames tricky statements. Going beyond 15 years has thin returns for GS because the pattern has drifted, though early starters building from scratch can skim 20 to 25 years for static fundamentals. What matters far more than the year count is analysing each paper properly.
Solving PYQs alone is not enough; analysing them is what clears Prelims. If you only tick options and check the answer key, you are missing the part that scores. The habit that works is taking each question apart until you know exactly why every wrong option is wrong, then tracing that theme back to your basic source. Pair decoded PYQ practice with limited standard sources and steady revision, and PYQs become the highest-return work in your whole prep.
Analyse UPSC PYQs by rebuilding the reasoning yourself instead of memorising the answer. Attempt the question cold and under time, note which options you eliminated and where you guessed, identify the syllabus theme and how often it has appeared, then go back to your basic source only for the exact gap the question exposed. Log the trap type, such as assertion-reason or matching pairs, so you recognise it next time. The goal is a method that transfers to a question you have never seen.
Yes, beginners should look at PYQs early, even before they feel ready. You will not solve many at first, and that is fine; the point is to see what the exam actually asks so your reading has direction. Use PYQs as a map: read a topic, then check how UPSC has questioned it, and let that set your depth. Beginners should pair this with a structured study plan so they build a base while learning what the real paper rewards.
For CSAT, the last 5 to 7 years of PYQs are enough. CSAT only needs the 33% qualifying mark and does not add to your merit rank, so the aim is to identify the question types you can solve reliably within the time limit and the ones worth skipping. Recent papers are the most representative of the current comprehension and reasoning style, so weight them heavily. If maths and reasoning scare you, cherry-pick the solvable questions rather than trying to crack every hard one.
Solve PYQs yearwise once your basics are in place, and use subjectwise practice earlier while you are still finishing a subject. Full timed papers show you each year cutoff and train you to judge, on the spot, what to attempt and what to leave. Subjectwise work is useful to test whether your coverage of a single area holds up. Most aspirants benefit from starting subjectwise, then shifting to yearwise papers as the exam nears.
You have done enough when you can sit an old paper cold and clear it comfortably, hitting roughly 80% on papers you have already analysed. The right signal is your hit-rate, not a target number of years. That 80% figure is a readiness barometer for you, not a claim about how many questions repeat in the exam. If you are consistently clearing analysed papers with time to spare, stop hoarding new material and shift your energy to revision and full-length mocks.
Solving PYQs is step one; decoding why each option is right or wrong is where the marks are. That decoding habit is the whole idea behind the Sherlocking test series.
Sources: The UPSC Civil Services (Preliminary) Examination has two papers, GS Paper 1 (counted for the cutoff) and CSAT Paper 2 (qualifying at 33%, not added to the merit rank). Confirm the current pattern against the latest official notification at upsc.gov.in . Year-range recommendations reflect the UnlockIAS mentor’s method, not an official rule; repeat-theme observations are his general experience, not a published statistic. Mentor quotes are reproduced verbatim from the UnlockIAS community archive.
Last updated: July 2026.