OCR A-level Computer Science (H446): papers, programming project and grade boundaries

OCR A-level Computer Science at a glance

The OCR course is a linear qualification with 100% terminal assessment: every component is completed at the end of the two years. It is regulated in England and has one exam series each year in May and June. Two papers are marked by OCR, and the third component is the programming project, which your teacher marks and OCR moderates.

That is 350 marks in total. Both papers are written, with no calculators. Both include synoptic assessment, and questions that test the quality of extended responses are marked with levels of response.

What each component covers

Component 01: Computer systems

This holds most of the content. It covers the characteristics of processors, input, output and storage (the fetch-decode-execute cycle, registers, pipelining, CISC and RISC, GPUs, Von Neumann and Harvard architectures); software and software development (operating systems, translators, the waterfall and agile lifecycles); exchanging data (compression, encryption, hashing, databases, networks, web technologies); data types, data structures and Boolean algebra; and legal, moral, cultural and ethical issues. Questions can ask you to follow and correct algorithms and programs as well as recall.

Component 02: Algorithms and programming

This is the problem-solving paper. It covers elements of computational thinking (abstraction, thinking ahead, procedurally, logically and concurrently), programming techniques and computational methods, and algorithms including standard searching and sorting, graph traversal and shortest path algorithms. Section A has shorter questions on writing algorithms, programming and problem solving. Section B begins with a scenario and asks mostly high-tariff questions about it, again on algorithms and programming.

Component 03 or 04: the programming project

You choose a computing problem and write a report on solving it, with analysis, design, development and evaluation, using a programming language from OCR’s accepted list. Your teacher marks it out of 70 against OCR’s criteria and OCR moderates a sample, either by online upload (03) or by post (04).

Pseudocode, calculators and what to expect

Learners are not permitted to use a scientific or graphical calculator in either exam, so practise binary, hexadecimal and floating-point arithmetic by hand. OCR sets code in its own pseudocode, shown in the specification. You do not have to memorise its syntax: you can answer in any style of pseudocode, or a programming language, as long as a competent programmer could work out what you mean.

Worked examples in OCR style

Data representation: two’s complement

Store the denary value −45 using eight bits. Start from +45 in binary, flip every bit, then add one:

Check it by adding the original: 00101101 + 11010011 gives 1 00000000, and the overflow bit is discarded, leaving 0. Practise number bases in our Number Systems course:

Algorithms: tracing a binary search

This function searches a sorted list in OCR-style pseudocode:

function binarySearch(list, target)
    low = 0
    high = list.length - 1
    while low <= high
        mid = (low + high) DIV 2
        if list[mid] == target then
            return mid
        elseif list[mid] < target then
            low = mid + 1
        else
            high = mid - 1
        endif
    endwhile
    return -1
endfunction

Trace it for the list [3, 8, 15, 21, 34, 42, 57] and the target 34. The list has 7 items, so the indexes run from 0 to 6.

Three passes find the item in a list of seven. Each pass halves the search space, so the time complexity is O(log n), compared with O(n) for a linear search. A trace table like this is one of the most common ways to earn marks in Component 02.

Complexity: reading nested loops

A loop over n items that contains another loop over n items does n × n steps, so it is O(n²). A single loop that halves the problem each time is O(log n). Say which loop dominates, because the question often asks for the time complexity of the whole routine.

Grades and June 2026 grade boundaries

OCR adds the three component marks into a total out of 350 and compares it with the boundaries. This table is taken from OCR’s document published on 14 September 2026 for the June 2026 series.

Worked through: 100 on Component 01, 95 on Component 02 and 55 on the project gives 250, which is a B (216) and 4 marks below an A (254). On the project alone, an A* boundary is 63 out of 70, so most of your marks there come from meeting every criterion fully.

Using the daily paper as a short timed mock

Uni-Sine publishes a free daily Computer Science Advanced paper, with a new one every day. It is a short timed paper of 8 questions worth 25 to 35 marks, marked instantly with worked answers. It is not an OCR paper and it is not the length of a 2 hour 30 minute exam, so treat it as quick practice, without a calculator, to match the real papers.

• It covers well: processor architecture, number representation and floating point, data structures, networking, databases and SQL, regular expressions, recursion, object oriented programming, searching, sorting and complexity, finite state machines and graph traversal.
• It does not cover: the programming project, or the long scenario questions of Component 02 Section B. Use OCR’s own past papers and specimen papers for those.

For advice on running a mock, see our A-level mocks guide, and for the whole subject the A-level Computer Science guide.

Taking the WJEC Eduqas course instead? Read our Eduqas A-level Computer Science guide, which has its own papers and boundaries.

Exam answers

Uni-Sine does not publish official answers or mark schemes for OCR A-level Computer Science papers. The answers and a worked mark scheme for each question come with today’s paper once you have submitted it and it has been marked, so you can see where every mark was won or lost.

Common mistakes in OCR A-level Computer Science

• Writing code that works but ignores the question. If it asks for a function that returns a value, return it, do not print it.
• Off-by-one errors. Check the first and last index of loops and the highest valid index of an array of n items, which is n − 1.
• Skipping the trace. If a question hands you an algorithm, work through it with a table before answering.
• Vague extended responses. For discussion questions, give points on both sides and link each to the scenario, then come to a conclusion.
• Leaving the project evaluation to the last week. The evaluation section is marked on its own criteria, and it needs testing evidence.

Read next

UK · A-levelOCR A-level Computer Science grade boundaries 2026OCR A-level Computer Science (H446) June 2026 grade boundaries, component by component, plus AS, with three years side by side.