A-Level Maths Exam Question Types: Pure, Mechanics & Statistics Explained

Eighteen years teaching A-Level Maths in Birmingham, and I still see the same pattern every summer. Students who struggled with the large data set questions in Paper 3. Students who lost three marks on a proof because they didn't state what they were trying to show. Students who could integrate perfectly in class but froze when the exam dressed it up as a rates-of-change context problem.

A-Level Maths is a specific exam with specific question types that reward specific habits. Once you know what those question types are and exactly what examiners are looking for in each one, the paper stops feeling unpredictable.

This guide covers the full range: Pure Mathematics, Mechanics, Statistics. I've written it primarily for AQA (specification 7357), Edexcel (9MA0) and OCR (H240), since those are the three boards sitting in the vast majority of UK sixth forms. Where there are meaningful differences between boards, I'll flag them.

How A-Level Maths papers are actually structured

Before getting into the question types, it's worth being precise about the format, because there's a lot of confusion about which papers contain what.

All three major boards run three 2-hour papers, each worth 33.3% of the A-Level. The split between Pure and Applied content varies slightly by board:

AQA 7357: Paper 1 is pure only (non-calculator). Papers 2 and 3 each contain pure material alongside either statistics or mechanics, with calculators permitted. The applied content isn't siloed: you can get a mechanics question mid-paper alongside integration.

Edexcel 9MA0: Paper 1 is pure non-calculator. Paper 2 is pure with a calculator. Paper 3 is the dedicated Statistics and Mechanics paper, split roughly 50/50. This makes Edexcel the most "modular-feeling" of the three for applied topics.

OCR H240: Paper 1 is pure non-calculator. Paper 2 adds a calculator. Paper 3 is Statistics and Mechanics. OCR is also notable for its explicit Problem Solving strand, with questions where the method isn't signposted for you.

All boards award grades A*-E. The grade boundaries shift year to year based on national performance, but a rough guide from recent series: A* typically requires around 80% overall; A around 68-72%; E is usually achievable around 40%.

Pure Mathematics: the largest slice

Pure content makes up roughly 67% of the marks across the three papers. It's the foundation, and also where most of the mark-scheme-unfriendly mistakes happen.

Proof questions

Proof appears on every A-Level paper, usually in the first few questions, which makes it dangerous because students rush it.

The two main types are proof by deduction (algebraic manipulation leading to a conclusion) and proof by contradiction (assuming the opposite, deriving a contradiction). Disproof by counter-example also appears, and it's where students give away marks unnecessarily by producing the counter-example without explaining why it disproves the statement.

What I tell every Year 13 class: write down what you're trying to prove before you start. It sounds obvious. It prevents at least one mark being dropped per sitting, in my experience.

A question I saw on an AQA paper a few years ago asked students to prove that the product of any two odd integers is odd. Half the class wrote a perfectly correct algebraic proof. About a third of them forgot to define their variables as integers at the start. Mark scheme said: "must state that m and n represent integers." Two marks lost on a question they'd essentially answered correctly.

Algebra and functions

This is the broadest Pure category. Expect questions on:

• Rational functions and partial fractions: split into partial fractions, then integrate or manipulate. AQA and Edexcel both favour this in the later questions of Paper 1.
• Composite and inverse functions: finding f⁻¹(x), domain restrictions, sketching f and f⁻¹ as reflections in y = x.
• Modulus functions: solving |ax + b| = cx + d by considering two cases. Students almost always miss one of the cases under exam pressure.
• Transformations of graphs: translating, stretching, reflecting. Marks are lost here not from not knowing the transformation but from not updating the coordinates of key points (intersections, asymptotes) on the sketch.

Calculus

Calculus questions are the most heavily weighted area of the Pure content, and the most varied.

Differentiation covers the chain rule, product rule, quotient rule, implicit differentiation and parametric differentiation. In my Year 13 class last autumn, I gave a mock containing a parametric differentiation question. A student who'd scored an A on every Pure test that year wrote dy/dt ÷ dx/dt correctly, got the right numerical answer, and then wrote "therefore the gradient is undefined" because the denominator was 0.1 rather than a whole number. Just nerves — the kind of mistake that costs A* candidates their grade.

Integration spans standard integrands, integration by parts, integration by substitution, and areas between curves. The multi-step integration questions on Edexcel Paper 2 in particular tend to embed integration inside a longer modelling problem, so you need to recognise you're integrating before you even pick a technique.

Differential equations involve forming and solving, including separation of variables. These typically appear as applied problems (population growth, Newton's law of cooling). The examiner wants to see the constant of integration retained until it's resolved using the boundary condition. Dropping it early is a common error.

Vectors

Vector questions in A-Level Pure are predominantly 3D: finding the angle between two vectors, finding the equation of a line, proving two lines intersect (or proving they're skew). The intersection proof question requires the working to be laid out line by line: set the two parametric equations equal, solve for the parameters, then substitute back to verify consistency.

OCR tends to ask "show that" in vector questions more than AQA or Edexcel. That phrasing means you must work towards the given answer, not just state it.

Exponentials, logarithms and sequences

Log equations and exponential modelling appear in most papers. The modelling version (usually disguised as "a colony of bacteria" or "radioactive decay") typically asks you to find a constant, then use the model to predict a value, then comment on the model's validity. That last part, commenting on validity, is worth 1-2 marks that many students simply leave blank.

Sequences include arithmetic, geometric and sigma notation. The binomial expansion has its own question type: expanding (1 + ax)ⁿ for large n, with a validity range to state. Students lose marks here for forgetting that |ax| < 1 needs to be written down as a condition.

Mechanics

Mechanics makes up roughly 16-17% of A-Level Maths marks. It appears on either Paper 2 or Paper 3 depending on the board, always alongside a calculator.

Kinematics

Kinematics questions come in two formats: constant acceleration (using the suvat equations) and variable acceleration (using calculus, differentiating displacement to get velocity, integrating acceleration to get velocity or displacement).

The suvat questions are deceptively straightforward and a fertile source of sign errors. Downward displacement, deceleration: students regularly define positive direction inconsistently mid-question. Set a positive direction at the start and stick to it.

Variable acceleration questions involving integration require the constant of integration. I cannot count the number of times I've seen a student integrate acceleration correctly to get velocity, then forget to apply the initial condition. The mark scheme will have a method mark for the integration and an accuracy mark for applying the boundary condition. Drop the second one and you're losing marks on work you've already done.

Forces and Newton's laws

The standard format: a particle on a slope, in a pulley system, or connected by a string. Draw a diagram. Resolve forces. Apply F = ma. That sequence, repeated carefully, picks up most of the available marks even when the geometry is awkward.

The friction questions are where marks get lost at A*-grade level. Students know F = μR, but applying it correctly when the particle is on the verge of moving (limiting equilibrium) versus actually moving requires reading the question carefully. The examiner says "the particle is on the point of moving" for a reason.

Moments questions typically involve a non-uniform plank or a rod in equilibrium. Taking moments about a sensible point (usually the pivot or one end, to eliminate an unknown reaction force) is the key technique. In an OCR paper from 2024, a moments question had four unknowns and a student needed to take moments about two different points plus resolve vertically. Students who attempted it in one equation had nowhere to go.

Projectile motion

Projectile questions split horizontal and vertical motion. Horizontal: constant velocity (no force acts horizontally). Vertical: constant acceleration under gravity (g = 9.8 ms⁻² on AQA and Edexcel; occasionally 9.81 ms⁻² in OCR contexts, worth checking your specification).

The multi-part structure usually goes: find the time of flight, find the range, find the maximum height, then (the discriminating part) find the angle of projection or initial speed given information about the landing point. That final part requires forming simultaneous equations from the two components and solving. Students who write the equations correctly but then reach for a calculator before eliminating one variable tend to create algebraic messes that cost time.

Statistics

Statistics is the component most students underestimate. It carries roughly 16-17% of total marks, and it has one feature unique to A-Level Maths: the large data set.

The large data set (LDS)

Every board provides a pre-release large data set that students can familiarise themselves with before the exam. AQA uses weather data from the Met Office, covering multiple UK weather stations over several years (temperature, rainfall, wind speed, humidity). Edexcel uses a similar meteorological dataset. OCR uses census-type data.

The LDS questions ask you to interpret, critique or calculate using data from this set. The examiner assumes you have seen the data before. Questions will reference specific station names, months, or values you'd only know if you'd read the data.

I've seen students answer LDS questions entirely in the abstract, treating them like generic statistics questions. That works for the calculation parts. It fails completely on the interpretation parts, which ask things like: "Give one reason why the mean wind speed in Camborne in January might not be a reliable estimate of the mean for the whole of the South West." That question requires you to know that Camborne is a coastal station with atypical exposure. You can't deduce that from the question alone.

Spend time with the LDS before the exam. Not memorising numbers, but understanding the dataset's structure, its limitations, and the geography of the weather stations involved.

Probability distributions

Binomial distribution appears on every paper. The pattern: identify n and p, calculate P(X = k) or P(X ≤ k), find the expected value (np), possibly comment on whether the binomial model is appropriate. That last part requires you to say something like "each trial must be independent," spelling out what independence means in the context of the specific scenario rather than just restating the textbook condition.

Normal distribution questions ask for probabilities, or for finding a value given a probability. They almost always require standardising to Z ~ N(0,1) and using either tables or your calculator's normal CDF. Know your calculator's statistical functions well: on Edexcel and AQA, the calculator CDF approach is fully credited in the mark scheme.

Hypothesis testing

Hypothesis testing is the Statistics question type students find hardest to set out correctly, despite the method being entirely procedural.

The steps are: state H₀ and H₁ in terms of the parameter; state the distribution under H₀; calculate the test statistic or the critical region; compare to the significance level; state a conclusion in context.

The conclusion must be in context. "Reject H₀" is worth one mark. "There is sufficient evidence at the 5% significance level to suggest that the mean weight of packages has decreased from 500g" is worth two. That extra mark is claimed in every sitting by students who've learned to write the full sentence. It takes about five seconds.

One-tailed vs two-tailed tests is a source of systematic errors. If the question says "test whether the proportion has changed" (unspecified direction), that's two-tailed: halve the significance level before comparing to the tail probability. Miss that, and every H₁ you've set up is wrong before you've done any calculation.

Correlation and regression

Correlation questions ask you to calculate or interpret Pearson's product-moment correlation coefficient (PMCC). The interpretation must reference the specific variables: not "there is strong positive correlation" but "there is strong positive correlation between temperature and ice cream sales." Regression questions ask you to use the equation of the regression line (given or calculated) to predict a value, then comment on whether the prediction is reliable, weighing whether the prediction falls within the data range (interpolation) or outside it (extrapolation).

Exam strategy: how to work the papers

Three 2-hour papers, each out of 100 marks. That's roughly 72 seconds per mark, useful as a mental check. If a 3-mark question has taken you 12 minutes, something has gone wrong.

My practical advice after 18 years of watching students in these papers:

Read the whole paper in the first five minutes. Not to answer, but to locate the questions you can do most fluently. Starting with a 10-mark calculus question you're confident in beats getting stuck on a 3-mark proof and burning time at the front of the paper.

On proof questions, state the assumption or what you're proving before line 1 of algebra. One sentence. It costs nothing and the mark scheme almost always rewards it.

On the large data set question, write a named reference to the dataset ("in the AQA Met Office data, Camborne is..."). Generic answers don't pick up the interpretation marks.

Show every step of working. The mark scheme on A-Level Maths awards method marks independently of correct answers. A calculation error on line 3 of a 5-line integration doesn't have to cost you all 5 marks — it costs you the accuracy marks but not the method marks, provided your working is visible and logically coherent. A blank page, or a final answer with no working, can only receive A marks. You'd be leaving method marks on the table.

If you want to structure your full revision period around these question types, our guide on GCSE to A-Level maths transition covers the broader preparation framework, and EduBoost's A-Level maths practice lets you drill specific question types with instant marking.

Questions I get asked every year

Which paper is hardest?

In my experience: Paper 1 for most students, because it's non-calculator and the pure content is dense. But "hardest" is personal. If statistics is your weak point, Paper 3 (or the applied paper on your board) is where you're most vulnerable. Identify that early and target it, rather than spending revision time on the content you already find manageable.

Do I need to memorise formulae?

Some are given in the formula booklet; some aren't. The quadratic formula is given. The product rule is not. For AQA, the binomial expansion for positive integer n is not in the booklet. Know what's in there and what isn't. It changes how you approach a question under pressure.

How should I use past papers?

First, use past paper questions by topic (just the integration questions from five years of papers, back to back). Once you're solid topic by topic, sit full papers under timed conditions. Mark them using the official mark scheme, not just checking final answers. Reading the mark scheme teaches you the language of A-marks and M-marks and what examiners consider a "valid method."

What grade do I need for my university course?

That depends entirely on the institution and course. Most Russell Group engineering degrees want an A or A*. Mathematics degrees commonly ask for A*A*A. Some economics programmes at top universities specify A-Level Maths at grade A as a minimum. Check UCAS entry requirements directly for each course you're applying to.

One final thought

The students I've taught who surprised themselves in A-Level Maths were rarely the ones who discovered a revision trick in the final week. They were the ones who, in October of Year 12, sat down and worked out what the exam actually looks like and built their preparation around that structure.

The question types in this guide don't change dramatically year to year. The large data set rotates; the individual problems are new; but the underlying categories (proof, calculus, vectors, kinematics, forces, hypothesis testing) show up reliably. Know them, know the marks available in each, and you walk into that exam room with a map rather than a mystery.

For targeted A-Level Maths practice aligned to your specific board, EduBoost's subject help section covers all three papers with worked examples and instant feedback.

---

David Crawshaw has taught A-Level Maths at secondary schools in Birmingham and Warwickshire for 18 years. He holds a BSc in Mathematics from the University of Nottingham and has marked A-Level papers for AQA.