Sign in
Cambridge IGCSE

Computer Science 0478

Data representation, hardware, networks and security, plus algorithms, programming, databases and Boolean logic.

Syllabus version: 2026–2028 syllabusChecked against the Cambridge syllabusExam year 2027 is covered by the 2026–2028 syllabus.
Subject review complete (20 Oct 2026). All 40 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: 21 approved as written, 19 corrected.
Paper 1 Computer Systems: 50%Paper 2 Algorithms, Programming and Logic: 50%Weighting2 papers
  • Paper 1 Computer Systems50%
  • Paper 2 Algorithms, Programming and Logic50%
Original questions40
Topics covered10/10
Papers2
Mock papers2

What is assessed

2 components. Question formats: short answer, structured, programming.

Paper 1 Computer Systems
1 h 45 min75 marksWeighting 50%

Compulsory short-answer and structured questions on Topics 1–6 (data representation, transmission, hardware, software, the internet, automated and emerging technologies).

What the paper looks like

About 9–11 compulsory questions, each built around a short context (a device, a network, a company) and split into parts worth 1–6 marks. Expect tick-box and term-completion parts, binary/hex/two’s complement conversions, shifts and file-size working without a calculator, and State/Describe/Explain questions on hardware, networks, security and AI. Some questions ask for an annotated diagram (e.g. the fetch stage or a DDoS attack). Marks reward precise technical terms applied to the context.

Short answerStructuredNo calculators; no additional materials
Checked against the Cambridge syllabus
Paper 2 Algorithms, Programming and Logic
1 h 45 min75 marksWeighting 50%

Compulsory short-answer and structured questions on Topics 7–10, ending with a 15-mark unseen scenario answered in pseudocode, Python, Visual Basic or Java.

What the paper looks like

About 9–11 compulsory questions in Cambridge pseudocode style: a 1-mark tick box or symbol match, data types, line-numbered algorithms to find errors in, CASE/IF/loops and string, file and array tasks to write, a flowchart or pseudocode trace table, a database table with primary key, data types and SQL to read or complete, and a logic circuit with truth table and expression. The paper ends with a 15-mark scenario (about 30 minutes) answered in pseudocode, Python, Visual Basic or Java, marked by levels on meeting the requirements, comments and suitable input/output messages.

Short answerStructuredProgrammingNo calculators; no additional materials
Checked against the Cambridge syllabus

Assessment overview

From the 2026–2028 syllabus. Checked against the Cambridge syllabus

All candidates
Paper 1Paper 2
Untiered: everyone takes Paper 1 and Paper 2 (each 50%); grades A* to G available.
Assessment objectiveIGCSE weighting
AO1 Demonstrate knowledge and understanding of the principles and concepts of computer science40%
AO2 Apply knowledge and understanding to a given context, including analysis and design of computational or programming problems40%
AO3 Provide solutions to problems by evaluating computer systems, making reasoned judgements and presenting conclusions20%
Changes in this syllabus version
  • Same two-paper structure as 2023–2025: Paper 1 Computer Systems and Paper 2 Algorithms, Programming and Logic, each 1 hour 45 minutes, 75 marks, 50%.
  • Two’s complement outcome reworded: convert positive and negative 8-bit integers to and from two’s complement.
  • File-size calculations must use 1024 (not 1000) for kibibytes, mebibytes and larger units.
  • More examples of magnetic, optical and solid-state storage; FDE abbreviation added to the CPU outcome.
  • Verification outcome now includes the purpose of visual and double entry checks and writing algorithms for them.
  • SQL list extended to ORDER BY ascending/descending and the logical operators AND and OR.
  • Clarified (v2, 2024) that Paper 2 coding answers must be in pseudocode, except the 15-mark scenario, which may use pseudocode, Python, Visual Basic or Java.
  • Later updates: operator names, up to three parameters for procedures and functions (v4, 2025) and the ROUND function description (v6, 2026).
Command words used in Computer Science papers (13)
Calculate
Work out an answer from the facts or figures given
Compare
Point out similarities and/or differences
Define
Give the exact meaning
Demonstrate
Show how something works or give an example
Describe
Set out the key points or features of something
Evaluate
Judge the quality, importance or value of something
Explain
Give reasons or purposes, making links clear and backing them up
Give
Produce an answer from a source or from memory
Identify
Name, select or recognise
Outline
Set out the main points briefly
Show (that)
Give structured working that leads to the stated result
State
Say clearly and briefly
Suggest
Apply knowledge to an unfamiliar situation where several answers are valid
Examiner guidance from recent reports (14)
Data representation. In two’s complement the leftmost bit is worth −128. Do not convert the positive value and simply set the first bit to 1; invert the bits and add 1 instead.
Data representation. A logical left shift of two places multiplies by 4, not 2. Show carries in binary addition when working is asked for, and use 1024 (never 1000) in file-size calculations.
Data transmission. Serial transmission sends bits one at a time down a single wire; it is not about whole packets taking turns. Use precise language: the same key encrypts and decrypts in symmetric encryption.
Data transmission. When asked how parity and ARQ work together, describe each step in order: count the 1s, compare with the agreed parity, send a negative acknowledgement or wait for a timeout, then retransmit.
Hardware. Describe the fetch stage register by register: PC to MAR along the address bus, instruction back along the data bus into the MDR, then into the CIR, and the PC is incremented. Do not just list fetch, decode, execute.
Hardware. Make answers fit the device in the question. A built-in sensor or storage device must make sense for that device, and “it is faster” earns nothing unless you say what is faster.
Software. The purpose of firmware is to provide low-level control of the hardware, for example starting the computer with a bootloader. Answers like “it runs the computer” are too vague.
The internet and its uses. Encryption does not stop data being stolen; it makes intercepted data meaningless. For firewalls, explain that traffic is checked against set rules and anything that breaks them is blocked.
Automated and emerging technologies. Apply sensors, microprocessor and actuator steps to the scenario: what is measured, what it is compared with, and what the actuator does. Machine learning answers must say how the system adapts in that context.
Algorithm design and problem-solving. Extreme data is the largest and smallest acceptable values; boundary data pairs each limit with the value just outside it. Explain an algorithm’s processes (input, process, output) when asked, not just its purpose.
Programming. Use the syllabus pseudocode exactly: CALL a procedure with its arguments, use OPENFILE … FOR READ/WRITE, READFILE, WRITEFILE and CLOSEFILE, and never use reserved words as identifiers.
Programming. In the 15-mark scenario, read the whole task first, use the arrays and variables given, initialise totals, validate inputs, comment your code and give messages with every input and output. Plan for ties when finding a maximum.
Databases. A primary key must be unique for every record, so sometimes no given field is suitable. Put text values in quotation marks in WHERE clauses, and use SUM for totals but COUNT for the number of records.
Boolean logic. Draw each gate with the correct symbol (XOR has the extra curved line) and never simplify the expression. Write a logic expression without “X =” in front unless asked, and work out intermediate columns in the working space.

Topic map

Practice types

Free sample questions

A podcaster increases the sample resolution used to record her voice. Which statement is correct?
A podcaster increases the sample resolution used to record her voice. Which statement is correct?
  1. More samples are taken each second, so the file is larger.
  2. The recording becomes less accurate but the file is smaller.
  3. More bits are used for each sample, so the recording is more accurate and the file is larger.
  4. The number of pixels increases, so the quality improves.

Answer: C. More bits are used for each sample, so the recording is more accurate and the file is larger.

Sample resolution is the number of bits per sample. More bits means each sample’s amplitude is stored more precisely, so accuracy and file size both increase. Option A describes sample rate; option D confuses sound with image resolution.

A system uses even parity. Which received byte must contain an error?
A system uses even parity. Which received byte must contain an error?
  1. 10110100
  2. 01100011
  3. 11010111
  4. 01011011

Answer: D. 01011011

With even parity every byte must contain an even number of 1s. 10110100 has 4, 01100011 has 4, 11010111 has 6, but 01011011 has 5 (odd), so it must have been corrupted.

FAQs

Are these official 0478 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.

Is a calculator allowed?

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