Cambridge International A Level

Computer Science 9618

Theory fundamentals and problem-solving with programming at AS; advanced theory, AI and further programming at A Level.

Exam year

Syllabus version: 2027–2029 syllabusChecked against the Cambridge syllabusExam year 2027 is covered by the 2027–2029 syllabus.

Subject review complete (19 Oct 2026). All 36 Computer Science questions were re-solved and checked against the syllabus and Cambridge mark-scheme conventions, and their auto-marking was tested with correct and incorrect answers: 12 approved as written, 24 corrected.

Paper 1 Theory Fundamentals25%
Paper 2 Fundamental Problem-solving and Programming Skills25%
Paper 3 Advanced Theory25%
Paper 4 Practical25%

Original questions36

Topics covered19/20

Papers4

Mock papers4

What is assessed

4 components. Question formats: structured, short answer, pseudocode, practical, programming.

Paper 1 Theory FundamentalsAS & A Level

1 h 30 min75 marks25% of A Level

Written paper on syllabus sections 1–8 (information representation to databases); answer all questions, mixing short and longer responses.

What the paper looks like

About 8 multi-part structured questions totalling 75 marks in 1 h 30 min, each built around a short scenario (a device, a small network, a database). Part marks mostly 1–4, occasionally 5–6 for explanations. Typical tasks: convert numbers, complete truth tables or draw circuits from expressions, trace assembly code in a given instruction set, write SQL, and explain/justify choices in context. Command words: state, identify, describe, explain, complete, write, justify. No calculator, so arithmetic is designed to be done by hand.

StructuredShort answerNo calculator

Checked against the Cambridge syllabus

Paper 2 Fundamental Problem-solving and Programming SkillsAS & A Level

2 h75 marks25% of A Level

Written paper on sections 9–12; answers in pseudocode, scenario-based problem solving, no program code required.

What the paper looks like

About 7–8 scenario-driven questions (75 marks, 2 hours) answered in Cambridge pseudocode, with an insert of built-in functions (string, numeric, date). Includes identifier tables, trace tables, flowchart-to-pseudocode, test data, structure charts and state-transition diagrams, then longer questions (6–8 marks) asking for a complete procedure or function. Marks are given for individual algorithm features (loops, conditions, return values, file handling), so partial solutions score.

StructuredPseudocodeInsert listing pseudocode built-in functions and operators

Checked against the Cambridge syllabus

Paper 3 Advanced TheoryA Level only

1 h 30 min75 marks25% of A Level

Written paper on sections 13–20 (A Level theory, including AI, algorithms and paradigms); answer all questions.

What the paper looks like

About 10–12 questions (75 marks, 1 h 30 min) covering A Level theory: floating-point conversions with working, protocols, Boolean algebra/K-maps and flip-flops, scheduling and memory management, encryption and certificates, AI graph searches, recursion tracing, ADT algorithms, OOP terminology and declarative facts/rules. Part marks typically 2–4; working must be shown for calculations. Pseudocode answers should follow the pseudocode guide.

StructuredShort answerNo calculator

Checked against the Cambridge syllabus

Paper 4 PracticalA Level only

2 h 30 min75 marks25% of A Level

On-screen programming in Java, Visual Basic or Python (console mode) on sections 19–20 (excluding low-level and declarative); code and test evidence pasted into an evidence document.

What the paper looks like

Three substantial practical tasks (75 marks, 2 h 30 min) on a computer using Java, Python or Visual Basic in console mode. Tasks are split into many small parts (1–7 marks): declare arrays or classes, read from supplied text files, implement sorts/searches, stacks/queues/linked lists/binary trees, recursion, OOP with inheritance and exception handling, then run tests. All code and screenshots of output go into the evidence document; anything missing from it scores zero.

PracticalProgrammingCentre computer without internet or email; candidate source files (no binary files) and an evidence document

Checked against the Cambridge syllabus

Assessment overview

From the 2027–2029 syllabus. Checked against the Cambridge syllabus

A Level

Paper 1Paper 2Paper 3Paper 4

All four papers in one series, or staged: AS (Papers 1 and 2) in year 1, then Papers 3 and 4 in year 2 with the AS result carried forward.

Assessment objective	A Level weighting
AO1 Knowledge and understanding of computer science principles and concepts	30%
AO2 Application of knowledge, including analysing problems in computational terms	30%
AO3 Design, program and evaluate computer systems, making reasoned judgements	40%

Changes in this syllabus version

No significant changes that affect teaching compared with the 2026 syllabus; subject content and assessment structure are the same.
Version 2 (December 2025) clarifies which papers are taken at each level and confirms calculators are not allowed in any paper.
Textbooks endorsed for examination from 2021 remain suitable.

Command words used in Computer Science papers (27)

Analyse: examine closely to show meaning, picking out elements and how they relate
Assess: reach a reasoned, informed judgement
Calculate: work out a value from the facts or figures given
Comment: give a reasoned opinion
Compare: point out similarities and/or differences
Complete: fill in the missing parts of a diagram or table
Consider: look at the information given and respond to it
Contrast: point out differences only
Define: give an exact meaning
Demonstrate: show how, or give an example
Describe: state the main points or characteristics of something
Develop: build on given information or take it further
Discuss: write in depth about an issue in a structured way
Draw: link a term to its description with a line
Evaluate: judge or calculate the quality, importance, amount or value of something
Examine: look into closely and in detail
Explain: give reasons or purposes, make relationships clear, say why and/or how with support
Give: produce an answer from a source or from memory
Identify: name, select or recognise
Justify: back up a case with evidence or argument
Outline: set out the main points briefly
Predict: say what is likely to happen using the information available
Sketch: make a simple freehand drawing showing key features in proportion
State: express clearly and briefly
Suggest: apply knowledge to propose ideas where several valid answers exist
Summarise: present the main points without detail
Write: produce an answer in a specified form, e.g. pseudocode or SQL

Examiner guidance from recent reports (14)

Information representation. Show every step of a base conversion or binary addition and clearly mark the final answer. In file-size questions keep track of bits versus bytes and kibi (1024) versus kilo (1000); say "the file takes less time to transmit", not "transmits faster", when discussing compression.

Communication. Answers about networks must be tied to the scenario: a generic list of star-topology benefits earns little unless it explains why it suits the organisation described. Know the exact sequence of CSMA/CD (listen, transmit, detect, jam signal, random back-off).

Hardware. A buffer is only memory, so it cannot "send" anything; say the processor transfers data from the buffer. For embedded or monitoring questions, refer to the specific device in the question rather than reciting a textbook definition.

Processor Fundamentals. Register transfer notation must use square brackets correctly: MAR ← [PC], then MDR ← [[MAR]]. When tracing assembly code, write a new row each time a value changes and track IX carefully with indexed addressing.

System Software. When comparing compilers and interpreters, be precise about who does what: the interpreter stops at an error, the programmer fixes it. Name OS tasks and then describe what the OS actually does for each one.

Security, privacy and data integrity. Do not repeat an example already given in the question stem. Distinguish validation (data is sensible) from verification (data matches the source or was not changed in transfer); parity and checksums are verification during transfer.

Databases. SQL answers are marked line by line: include the table names in INNER JOIN ... ON conditions, put aggregate functions with GROUP BY, and use the exact field names given. For 3NF explanations, refer to specific non-key attributes and their dependencies.

Algorithm Design and Problem-solving. Use only the functions and operators listed in the insert. Even long pseudocode questions have accessible marks for declaring variables, initialising totals and writing a correctly terminated loop, so always attempt them.

Data Types and Structures. When searching an array, stop the loop once the item is found and handle the not-found case explicitly. For stack and queue questions, update the pointer and check for full/empty before every push or pop.

Programming. Watch for quotation marks in trace and data-type questions: a value in quotes is a STRING, not a number or Boolean. With BYREF parameters, the change to the formal parameter alters the original variable; BYVAL changes are lost on return.

Software Development. Test data must match the stated rule: extreme data is the largest and smallest valid values, boundary data includes the first invalid values either side. Name the maintenance type and link it to the reason given in the scenario.

Data Representation (A Level). For floating-point conversions, show the mantissa and exponent working separately; a normalised positive mantissa starts 01 and a negative one starts 10. Use the correct data types in record declarations (telephone numbers are strings).

Computational thinking and Problem-solving. If an iterative algorithm is requested, a recursive one does not meet the requirement (and vice versa). State binary search preconditions (sorted data) and justify Big O answers by describing how the work grows with n.

Further Programming. In OOP answers, declare attributes as PRIVATE and access them through getters and setters, and call the parent constructor with SUPER.NEW in a subclass. In Paper 4, paste code as text rather than screenshots and include test evidence showing the inputs used.

Topic map

Practice types

Free sample questions

FAQs

Are these official 9618 past-paper questions?

No. They are original questions written to match the Computer Science syllabus topics. Official past papers are available to schools through Cambridge.

Why choose an exam year?

Cambridge updates syllabuses periodically. Your exam year selects the matching configuration so topics and components follow the right version.

How do AS and A Level differ here?

Both pathways share code 9618. AS pages include AS topics and components only; A Level pages add the A Level content and papers.

Is a calculator allowed?

It depends on the paper. Each component card shows the calculator rule from this configuration.