A laptop screen showing lines of program code in a dark code editor
Photo: Arnold Francisca / Unsplash

This article is published by Revision Library. Our own resource is described separately below.

How this list was made

Official board archives were ranked first because they are the source for whole papers and mark schemes. Reviewed on 17 September 2026, the article was checked against the board entries already used for Revision Library's past-paper research and the A Level Computer Science practice routes on this site.

At a glance

SiteBest forPriceAccount
AQABest for official past papers straight from the exam boardFreeNo account
Pearson EdexcelBest for bulk downloads by exam seriesFreeFree to browse; account for some tools
OCRBest for a past-paper finder by qualification and levelFreeFree to browse; account for some tools
WJEC / EduqasBest for building custom papers from real past questionsFreeNo account

The full comparison

AQA

Best for official past papers straight from the exam board

AQA is one of the UK's exam boards, and its site is the primary source for its own GCSE and A Level past papers, mark schemes and examiner reports, free to search and download without an account.

Use it for whole-paper timing and official wording before switching to topic practice.

Pearson Edexcel

Best for bulk downloads by exam series

Pearson Edexcel is one of the UK's exam boards. Its site lets anyone search and download most past papers, mark schemes and examiner reports for free; the most recent exam series is restricted to registered exam centres for a period after each sitting.

Its archive is strongest for full-paper rehearsal, while project work and targeted written weaknesses need a separate plan.

OCR

Best for a past-paper finder by qualification and level

OCR is one of the UK's exam boards, with a past-paper finder covering its GCSE and A Level qualifications, free to use. Some recent or secure materials are restricted under OCR's own past-paper release policy.

Use those papers to test timing and command-word discipline, then repair repeated algorithm, data or systems errors with focused practice.

WJEC / Eduqas

Best for building custom papers from real past questions

WJEC is a UK exam board; Eduqas is its England-facing brand. Past papers and mark schemes are made available to the public as well as to registered centres, free to download, alongside a distinctive free Question Bank tool for building custom practice papers from past questions with mark schemes and examiner's comments.

The Question Bank is helpful for written Computer Science practice, although students should still use their own board's full papers for final rehearsal.

Our own resource

Revision Library

Best for exam-style question sheets by topic

Computer Science students often practise by typing, which gives fast feedback and hides some of the written-paper problem. In an exam paper, the answer has to be reasoned on the page: trace the algorithm, show the conversion, compare the architecture choice or explain why a network decision matters without an editor correcting the next line. Those habits line up with the topic guides to searching and sorting algorithms, data representation and computer architecture and the LMC.

The A Level Computer Science library is built for that written side. It should sit after the board archive, not in front of it.

Use the A Level Computer Science topic guides when the concept needs rebuilding, then move into A Level Computer Science practice papers or Computer Science mock papers when mixed written work is needed. The project still needs its own planning, coding, testing and teacher feedback alongside this.

For shorter sessions, Computer Science flashcards can keep terminology active, and Computer Science revision formats can turn a remembered definition into a trace table, explanation or comparison. When the student needs mixed questions, Computer Science past-paper practice and Computer Science past-paper packs are the better routes.

A processor has a clock speed of 2.5 GHz, and each instruction takes an average of 5 clock cycles to complete. Calculate the number of instructions the processor can execute in one second.

Convert the clock speed to hertz: 2.5 GHz = 2.5 x 10^9 cycles per second.

Divide the cycles per second by the number of cycles needed per instruction: 2.5 x 10^9 / 5.

Calculate the result: 2.5 x 10^9 / 5 = 5 x 10^8.

Final answer: the processor can execute 5 x 10^8 (500 million) instructions per second.

Use the conversion and division in the A1.1 processor guide to check the method, then choose the A1.1 processor practice pack for a targeted written question. This calculation uses the supplied average cycles per instruction; clock speed alone does not measure processor performance.

Database normalisation worked example: a fictional community workshop stores BOOKING_RECORD(MemberID, MemberName, WorkshopID, WorkshopTitle, Seats). Each member has one stored name and can book several workshops; each workshop has one stored title and can have several members. A member has at most one booking per workshop, with a positive whole-number Seats value for that booking. Names and titles need not be unique. These business rules supply the dependencies; do not infer extra dependencies from the sample values.

The small atomic table has columns MemberID | MemberName | WorkshopID | WorkshopTitle | Seats, with rows M1 | Nia | W1 | Mending | 2; M1 | Nia | W2 | Printing | 1; M2 | Owen | W1 | Mending | 1. Each cell holds one value, with no repeating groups. The composite primary key is (MemberID, WorkshopID). Identify the highest normal form, explain an update anomaly and split the table while retaining each booking and its seat count.

The relevant functional dependencies are MemberID -> MemberName; WorkshopID -> WorkshopTitle; and (MemberID, WorkshopID) -> Seats. Together they imply that the whole composite key determines every column. Assume no other non-trivial dependencies except those implied by these rules. Neither ID alone identifies a booking: M1 books W1 and W2, while W1 has bookings from M1 and M2. Seats depends on the pair, not on either ID alone.

Every cell is atomic, so the table is in first normal form (1NF). MemberName depends on MemberID, only part of the composite key, and WorkshopTitle depends on WorkshopID, the other part. These are partial dependencies, so the table is not in second normal form (2NF). Update anomaly: if W1 is renamed Stitching, both W1 rows must change. Changing only the M1 row leaves W1 called Stitching for M1 but Mending for M2.

Final answer: split into MEMBER(MemberID, MemberName), primary key MemberID; WORKSHOP(WorkshopID, WorkshopTitle), primary key WorkshopID; and BOOKING(MemberID, WorkshopID, Seats), composite primary key (MemberID, WorkshopID). BOOKING.MemberID is a foreign key referencing MEMBER.MemberID, and BOOKING.WorkshopID references WORKSHOP.WorkshopID. With the stated dependencies, each non-key attribute now depends on its whole key and there are no transitive dependencies, so all three relations meet third normal form (3NF).

MEMBER contains (M1, Nia) and (M2, Owen); WORKSHOP contains (W1, Mending) and (W2, Printing); BOOKING contains (M1, W1, 2), (M1, W2, 1) and (M2, W1, 1). MEMBER to BOOKING and WORKSHOP to BOOKING are each one-to-many relationships. BOOKING preserves the many-to-many relationship between members and workshops and stores the seats for each actual booking. Joining BOOKING to MEMBER on MemberID and to WORKSHOP on WorkshopID reconstructs exactly the original three rows: it does not invent an M2 booking for W2. The unique referenced keys give each booking exactly one matching name and title. W1 can now be renamed in one WORKSHOP row.

Mark scheme, criterion | award: atomic cells and 1NF | B1; both partial dependencies and failure of 2NF | B1; inconsistent W1 titles after a missed update | B1; three relations with their correct primary keys | M1; both foreign keys | B1; retained booking pairs and Seats, with a lossless join explanation | A1; 3NF justified using the stated dependencies | B1. Total: 7 marks. Use the A1.8 databases and normalisation guide to practise checking keys and dependencies before splitting a table.

Browse A Level Computer Science practice

In the numbers

packs
19
question pages
141

A Level Computer Science practice packs, each with a full mark scheme

How to choose

First, separate the written papers from the programming project. Past papers are excellent for written theory, algorithms, data representation and explanation, but they do not show whether the student's project planning, implementation evidence or testing record is on track.

Use the board archive for the course you sit when the aim is a timed paper. The boards covered here publish their own papers and mark schemes, so that is where paper structure, wording and marking language should come from.

Practise by hand as well as by coding. A program that runs can still leave weak written habits: missing trace-table steps, unclear variable changes, vague comparisons or an answer that explains the final output without showing the method.

Once the same weakness appears twice, stop collecting full papers for a moment. Use Computer Science topic practice to rebuild that area, then return to a mixed paper so the student has to recognise the topic without a title giving it away.

For broader context, the A Level Computer Science revision websites guide compares general revision support, while the GCSE Computer Science past-paper guide keeps the lower level separate.

For general paper technique, read the guides to A Level past-paper websites and how to use past papers effectively. This article is narrower: it is about A Level Computer Science board papers and written-paper follow-up.

Frequently asked questions

Are A Level Computer Science past papers enough revision?
No. They are essential for written-paper practice, but the programming project needs separate planning, development, testing and feedback. A student can be improving on papers while still needing more careful project work.
Should I use my exam board's papers first?
Yes. Start with the awarding body for the course you are sitting because that gives the official paper structure, mark scheme and wording. Other practice can help, but readiness should be checked against your own board.
Why do written Computer Science questions feel different from coding?
Typing code gives immediate feedback, while a written paper asks the student to show reasoning without that feedback. Trace tables, conversions, comparisons and explanations all need visible method on the page.
When should I use Revision Library for A Level Computer Science?
Use the A Level Computer Science library after a full paper shows a specific written weakness. It is useful for topic practice, mock papers and paper-style questions, while the board remains the official source for whole past papers.

Revision Library resources in this article