IB Mathematics HL Paper 3 is the part of AA HL and AI HL where students are most directly tested on sustained problem solving. The questions are extended, unfamiliar and designed to require connected reasoning rather than one isolated technique.
The current course generation
For students taking final assessment through November 2028, HL Paper 3 uses the current course generation. Under the revised course first assessed in May 2029, IB keeps Paper 3 but reduces it from 55 to 50 marks and from 1 hour 15 minutes to 1 hour for both AA HL and AI HL.
What Paper 3 demands
- Read a long unfamiliar context without rushing.
- Identify what earlier parts establish for later parts.
- Choose representations and tools rather than being told the method.
- Carry results forward carefully.
- Explain mathematical reasoning, not only calculator output.
- Recover when a later part depends on a result you could not fully obtain.
A better Paper 3 workflow
- Read the whole problem once for structure.
- Mark definitions, constraints and quantities introduced by the question.
- Treat early parts as information-generating steps, not isolated marks.
- Preserve exact or symbolic results where later parts may reuse them.
- If stuck, state useful relationships and continue where possible.
- Return at the end to check whether conclusions answer the context.
AA versus AI
AA Paper 3 tends to reward analytical and symbolic control within extended mathematical problems. AI Paper 3 places stronger emphasis on modelling, interpretation and technology-rich contexts. Both still require mathematical communication and connected reasoning.
For course-specific content, use the AA owner or AI owner.
Official source checked 26 September 2026: IB AA and AI curriculum updates and current HL Paper 3 materials.
World Mathematics route: return to the World Mathematics Atlas for the wider map across examinations, curricula, competitions, mathematical objects and university routes.
IB Mathematics HL Paper 3: train extended-response problem solving rather than topic prediction
Paper 3 rewards sustained mathematical investigation
The challenge is often a sequence of connected prompts around an unfamiliar situation. Students need to use earlier parts, notice patterns and maintain notation across a longer argument.
Read the whole problem architecture
Before solving, scan the subparts. Later questions can reveal the direction of earlier work, while earlier results are often intended tools rather than disposable answers.
Use given results intelligently
If a question says hence or otherwise, look for a way to reuse the preceding result. Re-deriving everything wastes time and may miss the intended connection.
Conjecture, test, then justify
Extended problems may invite pattern recognition. Test small cases to discover structure, but distinguish exploration from proof or general justification.
Technology can support investigation
Graphing, numerical experiments or regression may reveal behaviour where permitted. The final response should still communicate the mathematical conclusion and evidence.
Maintain exactness strategically
Keep exact forms when they preserve structure; approximate when the problem or interpretation requires it. Premature decimals can obscure relationships.
Write intermediate conclusions
After a long calculation, state what the result means for the problem. This creates checkpoints and reduces the chance of carrying an irrelevant value forward.
Recover after a stuck subpart
Use any stated result, proceed conditionally where possible and protect later marks. Extended-response skill includes not allowing one blockage to end the entire problem.
Review by decision points
After practice, identify where the problem changed direction: a substitution, graph observation, recurrence, parameter interpretation or proof step. Those decisions are more reusable than memorising the whole solution.
Build stamina progressively
Start with untimed extended problems, then partial timing, then full examination conditions. Speed should follow familiarity with sustained reasoning.
Common errors
Treating each subpart as unrelated, ignoring a supplied result, overusing technology without explanation and abandoning later parts after one failure are costly patterns.
Transfer
The learner is ready when an unfamiliar multi-part problem feels like a mathematical investigation to organise, not a sign that an unstudied chapter has appeared.