Core structure
The December 2026 Joint CSIR-UGC NET bulletin describes a computer-based objective test with one paper, three parts and no break. Total exam time is 180 minutes and the maximum score is 200 marks. The same broad A/B/C structure appears across the five subjects, but the question counts, maximum attempts and wrong-answer deductions are not interchangeable. Use the subject row in the current bulletin when planning attempts; an older Life Sciences chart or a generic UGC-NET guide is not a substitute.
Part A is the general aptitude component, including reasoning and quantitative or analytical tasks. Part B is closer to conventional subject knowledge. Part C tests higher-order application and analysis within the chosen discipline. All parts contribute to one combined subject score. The official June 2026 cutoff notice says there is no separate cutoff for Parts A, B and C and that cumulative marks are considered for the merit list. That is important for planning: Part A can support the total, but it is not a standalone pass/fail paper.
December 2026 subject comparison
All five rows total 200 marks, but the number of questions offered is larger than the maximum number of attempts. Across subjects, the overall attempt limits are 75 for Chemical Sciences, 75 for Earth Sciences, 75 for LS-BT, 60 for Mathematical Sciences and 55 for Physical Sciences. Do not add all available questions and assume you must answer each: the bulletin expressly caps attempts by part. If more answers are marked than allowed, consult the instructions for how the responses are counted; do not assume all extras will be scored.
The penalty pattern changes how a guess affects the score. For Chemical Sciences and LS-BT, each wrong answer loses one-quarter of the marks allocated to that question in every part. Earth Sciences applies a one-third penalty in Part C while A and B use one-quarter. Mathematical Sciences has no negative marking in Part C, although A and B have penalties. Physical Sciences uses a quarter-mark deduction in all three parts, but its questions carry different marks across B and C, which produces deductions of 0.875 and 1.25 respectively. Treat “negative marking is one-quarter” as incomplete unless it is tied to the correct part and marks-per-question row.
Reading the table for your subject
In Chemical Sciences, the candidate sees 20 A, 40 B and 60 C questions, but may attempt only 15, 35 and 25. A and B questions award 2 marks each; C awards 4. Wrong answers deduct 0.5, 0.5 and 1 marks. The paper therefore rewards selective use of the 25 C attempts rather than trying to answer every analytical item. The same attempt caps and mark allocation apply to LS-BT, but LS-BT has 50 Part B and 75 Part C questions available, so the candidate faces a larger choice set in those parts.
Earth, Atmospheric, Ocean and Planetary Sciences offers the largest pool in this table: 150 questions. Yet the maximum remains 75 attempts. A/B questions carry 2 marks and a 0.5 wrong-answer deduction. Part C questions carry 4 marks and a 1.32 deduction, which is 33% of the question’s value rather than 25%. A strategy calibrated on Chemical Sciences’ Part C penalty would understate the cost of an incorrect Earth Sciences response.
Mathematical Sciences has a different pattern again. It offers 120 questions and permits 60 attempts: 15 in A, 25 in B and 20 in C. A correct answer yields 2, 3 and 4.75 marks respectively; the wrong-answer deductions are 0.5 in A, 0.75 in B and zero in C. The lack of a Part C penalty is specific to the December 2026 Mathematical Sciences table. It should not be transferred to other subjects or to a later cycle without checking that cycle’s bulletin.
Physical Sciences offers 75 questions with a maximum of 55 attempts. The A/B/C marks are 2, 3.5 and 5; wrong-answer deductions are 0.5, 0.875 and 1.25. Its Part C questions have the highest value per correct response among these rows, but an incorrect answer also loses more marks in absolute terms. The useful decision is not simply “attempt the high-mark part”: consider how reliably you can solve that part under time pressure, and track accuracy at the actual subject-specific penalty.
Turning the scheme into an attempt plan
Build a mock-test score sheet with one row for each part. Record attempted, correct, incorrect and left-unattempted counts, then apply the correct subject’s mark values and penalties. This reveals whether missed marks come from time, concept gaps, excessive guessing or poor part selection. Keep the count within each maximum attempt limit. In a paper where several options are left blank, choosing the best-supported questions is more useful than filling the full allowance without regard to accuracy.
Use the official scheme to check any mock or previous-paper analysis. Some third-party summaries omit the number of questions offered, confuse “questions available” with “attempts allowed,” or round a fractional penalty. For Earth Sciences Part C, 1.32 is the listed deduction; for Mathematical Sciences Part C, the bulletin lists no penalty. Preserve the exact figure in practice calculations. These small differences matter when comparing a mock score with a qualifying threshold or deciding whether a guess had positive expected value.
The broad CSIR-HRDG exam-scheme page explains the three parts and 200-mark structure, but the December 2026 bulletin is the stronger source for the exact current-cycle table. If a later official notice revises a pattern, use the later notice for the affected cycle. The published December 2026 structure is not evidence for an unannounced 2027 pattern.