On this page
Numerical value questions have no options to eliminate and no way to work backwards from choices. You type a number, and it is either right or wrong. In JEE (Main) 2026 they also carry negative marking. This post sets out what NTA's bulletin says, then gives a method for answering them and a weekly routine to practise it.
How many numerical questions are there, and how are they marked?
NTA's Information Bulletin for JEE (Main) 2026 describes Paper 1 (B.E./B.Tech.) as 75 questions for 300 marks in three hours. Each of mathematics, physics and chemistry has two sections: Section A with 20 multiple-choice questions and Section B with 5 questions "whose answers are to be filled in as a numerical value". The pattern table counts all 75 questions towards the 300 marks and does not describe any optional choice within Section B.
The marking scheme for Section B in the bulletin is:
Correct answer: +4
Incorrect answer: −1
Unanswered or marked for review without a value: 0
The bulletin also says you must enter "the correct integer value of the answer using the mouse and the on-screen virtual numeric keypad", that "the answer should be rounded off to the nearest integer", and that candidates should do calculations "with the constants given (if any) in the questions". Calculators are not allowed, and all rough work goes on the rough sheets provided at the centre.
Patterns change between years, so read the bulletin for your own session on the JEE Main website when it comes out.
Should you guess a numerical answer?
Almost never. On a multiple-choice question, a random guess has a one-in-four chance. On a numerical question, the space of possible integers is huge, so a blind guess is close to certain to lose a mark. With +4 for right and −1 for wrong, a guess only breaks even if you are at least 20% sure of it (because 0.2 × 4 = 0.8 = 0.8 × 1). A half-finished calculation that gives you a likely value is different: if you trust the method and checked the units, enter it. Our post on negative marking and when to guess works through the same expected-value logic for multiple-choice questions.
How do you avoid rounding and unit errors?
Most lost marks in this section come after the physics or chemistry is already right. Three habits fix them.
1. Find the unit the answer must be in before you start. If the question asks for an energy "in kJ/mol", your final integer must be in kJ/mol.
Worked example 1. A plot of log k against 1/T has slope −4250 K. Find the activation energy in kJ/mol (R = 8.314 J/K/mol).
From log k = log A − Ea/(2.303RT), the slope is −Ea/(2.303R).
Ea = 4250 × 2.303 × 8.314 J/mol. First, 4250 × 2.303 = 9787.75. Then 9787.75 × 8.314 ≈ 81,375 J/mol.
In kJ/mol that is 81.375, so you would enter 81. Entering 81375 would be marked wrong.
2. Keep extra digits until the last line. Rounding at each step can push the final value across a boundary.
Worked example 2. A steel wire 0.72 m long has a mass of 5.0 × 10⁻³ kg and tension 60 N. Find the speed of transverse waves in m/s. NCERT works this in Chapter 14 Waves.
μ = 5.0 × 10⁻³ / 0.72 ≈ 6.94 × 10⁻³ kg/m.
v = √(T/μ) = √(60 / 6.94 × 10⁻³) = √8640 ≈ 93.0 m/s.
Enter 93. If you had first rounded μ to 7 × 10⁻³, you would get √8571 ≈ 92.6, which still rounds to 93 here, but only just.
3. Treat a value that lands near .5 as a warning. Because the answer must be entered as an integer, a result such as 12.49 or 12.51 sits right on the rounding boundary. Recheck the constants and the arithmetic before you decide which way it falls.
A five-line method for every Section B question
Underline what is asked and its unit.
Write the principle in one line before any algebra, for example "energy conserved, no friction".
Solve in symbols first, then substitute numbers once.
Check: dimensions, sign, and whether the size is physically sensible. Our dimensional analysis check takes ten seconds.
Round once, at the end, and enter the integer.
If you keep getting the method wrong rather than the arithmetic, the Fermi Problem Library has worked problems with a step-by-step solver that explains the reasoning, so you can compare your line 2 with the full logic of a solution.
When should you attempt Section B during the paper?
There is no single right order, but many students lose time by leaving all numerical questions to the end and then rushing them. A plan that works for many: in each subject, read all five Section B questions in your first pass and solve the ones whose method you see at once. Come back to the rest after Section A. Our guide to time management in the exam covers how to set these checkpoints.
A weekly practice routine
Monday: Ten numerical questions from one chapter. Write the unit of the answer before solving each.
Wednesday: Ten mixed questions from three chapters, with the five-line method written out in full.
Friday: One timed block of fifteen numerical questions (five per subject) in 30 minutes, no calculator.
Sunday: Go through every wrong answer and label it: wrong principle, wrong unit, arithmetic, or rounding. Spend next week's extra practice on the largest group.
Take one numerical question you got wrong in your last mock and redo it today with the five-line method, unit first.
