GATE ECE Subject-Wise Weightage, Counted from Five Papers

Across five GATE Electronics and Communication papers read one question at a time — 2010, 2019, 2020, 2025 and 2026 — General Aptitude is exactly 15 marks every time, and the largest technical section is Networks, Signals and Systems at a mean of 17.0 marks. Engineering Mathematics follows at 12.0, and Electromagnetics is the smallest at 7.4.

Those are counts checked against each year's official answer key, not figures from a published weightage table. The difference matters more than it sounds, and the second half of this guide shows how much.

Key takeaways

• No organising institute has ever published a marks breakdown for the EC paper. Every weightage table in circulation, including this one, is somebody's count.
• Five papers were counted question by question: 2010, 2019, 2020, 2025 and 2026. Each column closes at 65 questions and 100 marks, with General Aptitude at exactly 15.
• Networks, Signals and Systems is the largest technical section at a mean of 17.0 marks and Electromagnetics the smallest at 7.4.
• Small sections are not cheap sections. Control Systems asks 4.8 questions a paper at 1.67 marks each, the heaviest in the paper.
• Three published tables were checked and rejected: one reprints its own 2025 column as 2026, and four sites give four inconsistent figures for 2020 that all descend from one miscount.

The full counted table

Marks by section and year. The mean is the average of the five counted papers, and each is also that section's percentage share of the 100-mark paper.

The last row is six marks of the 2010 paper testing material the 2027 syllabus no longer contains. In the other four years that figure is zero, so only the oldest counted paper needs reading through a filter.

One inversion stops you over-reading the ranking. Networks tops the technical table in 2010, 2019, 2020 and 2026. In 2025 it does not: Engineering Mathematics took 15 marks and Networks 13. That is the only inversion in the counted record, and a reason to treat the top of the table as a block rather than an order.

How the counting was done

Each paper was taken in question order. Every question was read, assigned to one of the eight technical sections of the 2027 syllabus rather than the one in force in its own year, and recorded with its topic, format and mark value. Where a question honestly belonged to two sections, both were recorded; where it tested material the 2027 syllabus no longer contains, it went into the residual row.

Every year then had to pass four checks before its column was used:

1. 65 questions. Nothing dropped, nothing double-entered.
2. 100 marks. The section column sums to exactly 100.
3. General Aptitude exactly 15. The check that catches the most common failure in published tables.
4. Row-by-row agreement with that year's official answer key on mark value and format.

You can verify check 2 on the table above: every column sums to 100. A second check does not depend on the first. Every question is worth 1 or 2 marks, so 2-mark items in any group equal marks minus questions. General Aptitude across five papers is 50 questions and 75 marks, so 75 − 50 = 25 two-mark items, five per paper — the known structure of the section, set out in the General Aptitude preparation guide.

Questions, not just marks

Dividing each section's five-paper mark total by its question total gives the average weight of one question. Most weightage articles never print this, and it changes how you read the table above.

Control Systems asks the fewest questions of any section, 4.8 per paper, and charges the most for each: 1.67 marks, with two-thirds of its items worth 2 marks. Electronic Devices is second at 1.63, while Engineering Mathematics at 1.46 spreads its marks over more, cheaper questions. So a thin pass over everything works in Mathematics and fails in Control Systems, where four or five questions carry the section. Keep this in mind when you plan your attempt order and the on-screen calculator time numerical-answer questions cost you.

Why the published weightage tables were rejected

Five compilation sites were pulled and 478 raw marks observations recorded before any of it was interpreted. None of it is used — not because the sites are dishonest, but because checkable failures make them unusable as evidence.

A 2026 column that is a 2025 column

One site's published 2026 weightage is its own 2025 weightage reprinted unchanged in all nine sections, digit for digit, while that site's earlier rows differ from one another throughout. A row that reproduces the previous year across every section is a value carried forward, not a paper counted.

The counted papers are not the same in seven of those nine sections. Subtract the 2025 column from the 2026 column in the table above: Engineering Mathematics −6, Networks +7, Electronic Devices 0, Analog Circuits −1, Digital Circuits +4, Control Systems −3, Communications −2, Electromagnetics +1. The positive entries sum to 12 and the negative to −12, as they must, since both years total 100. Twelve marks moved out of four sections and into three between those two papers — one mark in every eight — and the published table records none of it.

That table tells a candidate Engineering Mathematics is a 15-mark section going into a paper where it was worth 9, and Networks a 13-mark section going into a paper worth 20. The two errors point in opposite directions and compound.

Four sites, four figures, one ancestor

Four separate sites give four different 2020 weightages. The published totals are 105, 90, 90 and 84. They are not four measurements. The shared core totals 90 where the paper's technical sections give 85, and the five-mark excess is locatable: Control Systems, published at 12, counted at 6. The 105 is that inflated core with General Aptitude added back; 90 is the same core with General Aptitude dropped; 84 is 90 with one cell misparsed.

One subject double-counted and one section dropped, propagated four ways and presented as agreement. Copying turns one error into four sources, and four sources look like corroboration.

Digital Circuits 2026: 13 against 7

The counted 2026 paper gives Digital Circuits 13 marks. The published table gives 7. Six marks is about four questions at Digital Circuits' own measured rate of 1.53 marks per question.

The misdirection is larger than the arithmetic, because it moves the section across a tier boundary. At 7 marks Digital Circuits reads as a small section, the kind a candidate under time pressure drops. At the counted 13 it was the second-largest technical section in the 2026 paper. It is also the most volatile section in the record, 6 marks in 2010 rising to 13 in 2026, which is why a stale table gets it most wrong.

The ten-second test

The diagnostic that catches all three failures applies to any weightage table, including this one: add the column. If it does not come to 100 with General Aptitude at 15, it is not a count of the paper. A column that does close can still be wrong, but only by trading marks between two sections, which is visible to anyone willing to read the two questions. A column that does not close was never reconciled at all.

What 2019 says, and what it cannot

The five counted papers were set by four institutes: IIT Guwahati in 2010 and 2026, IIT Madras in 2019, IIT Delhi in 2020 and IIT Roorkee in 2025. GATE 2027 is set by IIT Madras, which makes 2019 the only counted paper from the institute writing the next one. Confirm the institute and the schedule on the official site and in the GATE 2027 exam dates guide.

2019 is also the flattest paper counted. Every technical section in it landed between 9 and 14 marks, a spread of 5, against 9 in 2025 and 15 in 2010. But one paper is one paper: a single observation cannot separate an institute's house style from an ordinary year that came out flat, and there is no second IIT Madras paper in the record to test it against. What it supports is a negative — do not build a plan that requires a section to be skippable.

How to use these numbers

Study in the order the means give, not the order the syllabus prints. Networks and Engineering Mathematics mean 29 marks a paper between them and deserve the first block of your time. Analog Circuits, Communications and Digital Circuits are 30.6 marks and move the most between papers, which is where a stale table does the most damage. Electronic Devices, Control Systems and Electromagnetics are 24.2 marks: cheap to cover badly, expensive to skip.

Then treat the ranges as seriously as the means. Digital Circuits ran 6 to 13, Engineering Mathematics 9 to 15, Networks 13 to 21. A plan built on a mean alone is built on a paper nobody has set. The ECE book prints this table with a per-topic breakdown under each section and the 78 concepts that recur, ranked.

Five papers is five papers: exact and checkable is not the same as complete, and any single paper carries a fifth of the vote here. Once you have marks in hand, how the GATE score is calculated explains what they turn into.