CSIR NET Exam Pattern for Maths: What to Attempt and When

The Mathematical Sciences paper (subject code 704) is a computer-based test of 180 minutes in three parts, all objective questions, with no break between parts. As of October 2026, the scheme below is the one in section 3.2 of the December 2026 bulletin (printed page 18). Check the counts against each new bulletin when it is published.

Part A is general aptitude: logical reasoning, graphical analysis, numerical ability, series and puzzles. Part B asks conventional subject questions from the syllabus. Part C asks higher-order questions that apply concepts to a problem. The question paper is bilingual (Hindi and English), and the English version is final if the two differ. Our syllabus explainer covers what each part draws on.

The bulletin's dates at a glance (printed page 4) give 19, 22 and 27 December 2026 for the examination, in shifts of 9:00 to 12:00 and 15:00 to 18:00, and describe these dates as tentative. Your admit card shows your date and shift; see our CSIR NET exam date page for the schedule.

The other four papers (Chemical, Earth, Life Sciences, Physical) have their own counts and penalties in the same section of the bulletin. Do not use their numbers for Mathematical Sciences.

Five rules matter: Part C is all-or-nothing, the attempt limit is enforced, blanks score zero, you can move freely between sections, and you cannot use a calculator.

• Part C can have more than one correct option. The bulletin states that in Part C of Mathematical Sciences each question may have one or more correct options, and credit is given only when all of them are identified (section 3.2, printed page 19, and Annexure IX, printed page 51). Parts A and B have exactly one correct option. The CSIR Human Resource Development Group (HRDG) scheme adds that no credit is given if an incorrect option is marked as correct (CSIR HRDG scheme of examination). Ticking two of three correct options earns nothing.
• The limit is enforced. If you answer more than 15, 25 and 20 questions in Parts A, B and C, only the first 15, 25 and 20 are evaluated (section 3.2). Our advice is to keep a running tally and stop at the limit. A surplus answer cannot add marks and may be counted in place of a better one.
• Blanks score zero. Only questions marked "Answered" or "Answered and marked for review" are evaluated (Annexure IX, printed page 53).
• You may move between sections and questions at any time within the 180 minutes, so there is no section timer to fear (same page).
• A calculator is not permitted. Annexure IX (printed page 51) states that use of a calculator is not permitted, and the bulletin's list of barred items (printed page 31) includes calculators. Rough work goes on the sheets provided at the centre. Practise arithmetic by hand.

In Parts A and B a guess breaks even at 20% accuracy, and in Part C a guess never costs marks but still uses one of your 20 answer slots. The 20% figure is our working from the bulletin's marks.

The working: the expected gain of answering is p × marks for a right answer − (1 − p) × marks lost for a wrong answer, where p is your chance of being right. Setting that to zero gives p = loss ÷ (gain + loss).

• Part A: 0.5 ÷ (2 + 0.5) = 0.2, or 20%.
• Part B: 0.75 ÷ (3 + 0.75) = 0.2, or 20%.

The penalty in each part is exactly 25% of the marks for a correct answer (0.5 of 2, 0.75 of 3), which is why both break-even points match.

A blind guess among four options is right 25% of the time, so it is worth only +0.125 marks in Part A and +0.1875 in Part B on average, with a real chance of losing marks. Remove one option and your chance rises to about 33%. Our recommendation is to guess in Parts A and B only after eliminating at least one option, and to prefer a question you can solve over a guess, since the caps of 15 and 25 mean every guess takes a slot a solved question could have used.

Part C works differently. If any non-empty combination of the four options could be correct, there are 15 possible answers (2⁴ − 1), so a blind guess is right 1 time in 15, about 6.7%, worth about 0.32 marks (4.75 ÷ 15). That assumes every combination is equally likely, which is our model and not an official figure. The penalty is zero, but you only have 20 slots, so spend them on questions where you can verify each option.

In our illustrative model, extra wrong answers in Parts A and B cost only 2.5 marks. The bigger loss, 9.5 marks, comes from assuming guessing costs you two Part C answers. The scenarios below are our model, not data from real candidates.

Marks = 2 × A right − 0.5 × A wrong + 3 × B right − 0.75 × B wrong + 4.75 × C right.

Against the balanced row, the over-guessing row is 12 marks lower: 2.5 marks of extra penalties (7.25 against 4.75) plus 9.5 marks for the two Part C answers (2 × 4.75). Against the Part C heavy row the gap is 15.25 marks.

Part C is where an extra answer is worth most: 95 of the 200 marks, or 47.5%, come from only 20 questions, and 13 correct instead of 10 adds 14.25 marks (3 × 4.75).

The pattern sets a floor, not the cut-off: 33% of 200, which is 66 marks, for General, EWS (Economically Weaker Section) and OBC (Other Backward Classes) candidates, and 25%, which is 50 marks, for SC, ST and PwD/PwBD candidates (Scheduled Castes, Scheduled Tribes, and persons with disabilities or benchmark disabilities). The figures are in section 11.3 of the bulletin (printed page 32).

There is no separate cut-off for Part A, B or C; marks from all parts are added for the merit list. The bulletin sorts qualifiers into three categories (printed page 13): Junior Research Fellowship (JRF) with Assistant Professor eligibility, Assistant Professor with PhD admission, and PhD admission only. The results committee decides how many qualify in each, so reaching the floor does not guarantee a category. Our cut-off guide has a target-score worksheet.

As a sense of scale, Part C alone reaches the General floor with 14 correct answers (14 × 4.75 = 66.5) and the SC/ST/PwD floor with 11 (11 × 4.75 = 52.25). In our view, aiming for the floor is not a plan: JRF needs a score well above it, and our cut-off guide shows where recent cut-offs have fallen.

Split the 180 minutes by part, then log every mock test by part and check four numbers against fixed pass tests. The time split is our recommendation and not an official figure.

The total is 180 minutes. The average across all 60 attempts is 3.0 minutes (180 ÷ 60), but you read through 120 questions to choose them, so spend your first minutes marking the questions you can solve and leave the rest.

Our mock analysis guide shows what to look for in the results.

We teach CSIR NET Mathematics through live online classes and recorded courses, and we run a mathematics test series, including the Core Concept Diagnostic test series, designed by SB Sir (Sunil Bansal). SB Sir teaches Pure Mathematics, Eshita Mam teaches Applied Mathematics, and Vivek Sir teaches Abstract Algebra and Real Analysis.

Our courses include PDF notes, assignments, discussion of previous-year questions, mock tests and analytics. The app dashboard tracks your course completion and test scores. Record the right, wrong and blank counts for each part yourself using the log above. You can start with free material: demo classes, notes and previous-year questions.

As of October 2026, the CSIR-NET December 2026 Live Batch costs ₹11,999 one-time or three instalments of ₹3,999.67. The CSIR-NET and GATE Mathematics 2027 course is ₹15,999 one-time with access through 30 June 2027, or four instalments of ₹3,999.75. A ₹999 Demo Plan gives 10 days of access. You can see the options in our course store. If you want help choosing between live and recorded classes, or want to know which mock tests fit your stage, contact us.