Rabi and summer crops now give India 53% of its foodgrains, up from 36% in 1966-67, on land that grew by a third while kharif area stood still. Maize, rice and moong are moving into the winter and summer seasons, and each season’s harvest reaches the mandi in a narrow window, wheat mostly in a single month.
India’s Winter Harvest Now Outgrows Its Monsoon One. Kharif, Rabi and Zaid, Crop by Crop: Area, Yield and When It Reaches the Mandi
India grows its crops in three seasons. Kharif is sown with the south-west monsoon in June and July and harvested from September to December: rice, maize, bajra, jowar, soybean, groundnut, cotton, tur, moong and urad. Rabi is sown after the monsoon, from October to December, mostly on irrigated or residual-moisture land, and harvested from February to April: wheat, gram, mustard, lentil and winter maize. Zaid, which the agriculture ministry now reports as “summer”, fills the gap between March and June with irrigated rice, moong, maize, groundnut and vegetables.
This piece uses the area, production and yield estimates the Department of Agriculture publishes through its UPAG portal, by season and crop since 1966-67, and Agmarknet’s monthly mandi arrivals from January 2019 to October 2026, as compiled in this blog’s open india-mandi-arrivals repository. The latest year, 2025-26, is the ministry’s third advance estimates. Because summer crops were counted with rabi until 2022-23, long-run comparisons below put rabi and summer together.
The winter harvest has overtaken the monsoon one
In 1966-67 rabi and summer crops gave India 36% of its foodgrains. In 2025-26 they gave 53%. Rabi first produced more foodgrain than kharif in 2002-03, a drought year for kharif, and has done so in 15 of the 24 years since, including every year from 2017-18. The reason is not that kharif has shrunk in output; it is that kharif has not grown in land. Over the last twenty years, comparing three-year averages for 2003-06 and 2023-26, kharif foodgrain area was flat at about 733 lakh hectares, while rabi and summer foodgrain area grew +32% to 640 lakh hectares. Both seasons raised yields by about half.
| Group and season | Area, lakh ha | Yield, kg/ha | Production, lakh t |
|---|---|---|---|
| Foodgrains, kharif | 732 → 733 (+0%) | 1,497 → 2,279 (+52%) | 1,097 → 1,671 (+52%) |
| Foodgrains, rabi + summer | 485 → 640 (+32%) | 2,002 → 2,944 (+47%) | 970 → 1,884 (+94%) |
| Pulses, kharif | 112 → 117 (+4%) | 468 → 635 (+36%) | 52 → 74 (+42%) |
| Pulses, rabi + summer | 116 → 162 (+40%) | 735 → 1,129 (+54%) | 86 → 183 (+114%) |
| Oilseeds, kharif | 166 → 198 (+19%) | 956 → 1,330 (+39%) | 159 → 263 (+66%) |
| Oilseeds, rabi + summer | 98 → 107 (+10%) | 1,023 → 1,453 (+42%) | 100 → 156 (+56%) |
Rabi land is also more productive. In 2025-26 a hectare of rabi or summer foodgrains yielded 3,013 kg against 2,323 kg in kharif, mainly because most rabi crops are irrigated and grow in cooler weather with fewer pests. And rabi output is steadier: since 2005-06 the year-to-year change in kharif foodgrain production has had a standard deviation of 5.9%, against 4.3% for rabi. The worst kharif year was 2009-10, a drought year, when output fell 12.1%; the worst rabi year was 2014-15, when unseasonal rain and hail cut output 9.1%.
Pulses show the same pattern more strongly: rabi pulses (mainly gram and lentil, plus summer moong) gained +40% in area and +54% in yield, while kharif pulses gained +4% and +36%. Oilseeds are the exception, growing mostly in kharif, where soybean spread.
Crop by crop: where land moved, and where yields did the work
The biggest shift is maize moving into the winter. Rabi and summer maize area grew +358% in twenty years, from 8.1 to 37.0 lakh hectares, at yields of about 5,285 kg a hectare, against 2,910 for kharif maize. In 2025-26, Andhra Pradesh, Telangana and Bihar each grew about 4 million tonnes of rabi maize. Rice is following: rabi and summer rice area roughly doubled (+105%), led by West Bengal’s boro crop. Moong grown after the rabi harvest, much of it summer moong in Madhya Pradesh, grew +259% in area and +144% in yield.
In kharif, land moved out of millets and into soybean, cotton and maize. Kharif jowar area fell −68% and kharif bajra −30%, though bajra yields rose +56%, which kept output roughly level. Sesamum lost −43% of its area. Soybean is the cautionary case: its area rose +77% while its yield barely moved (+1%), so all its growth came from land. Kharif groundnut went the other way, losing −16% of its area but nearly doubling its yield. Wheat, the largest rabi crop, gained +24% in area and +35% in yield; gram gained +37% and +51%.
Zaid, the third season, is growing fastest
Only four years of separate summer data exist, but they show the season growing quickly. Summer foodgrain area rose from 62.0 lakh hectares in 2022-23 to 76.6 in 2025-26 (+24%), and production from 161 to 200 lakh tonnes. Summer pulses area went from 21.7 to 27.2 lakh hectares. Summer maize area doubled, to 10.6 lakh hectares, with Uttar Pradesh, West Bengal and Bihar growing most of it. Summer rice (32.6 lakh hectares) is concentrated in West Bengal, which grew about 5 million of the 11 million tonnes; summer moong is concentrated in Madhya Pradesh, about 1.8 million of 2.4 million tonnes. All of these crops depend on groundwater or canal water in the hottest months, which is why the season’s growth is as much a water question as a farm one.
When the harvest reaches the mandi
Mandi arrivals follow the harvest closely, and the calendar shows how concentrated each season’s selling is. Kharif crops arrive in a rush from October to December: 68% of a year’s paddy arrivals, 53% of soybean and 50% of groundnut land in those three months. Cotton follows from November to January (57%) and tur, the slowest kharif crop to mature, from January to March (53%). Rabi crops arrive from March to May: 45% of mustard in March and April, 51% of gram from March to May, and 39% of wheat in April alone (60% in April and May). Maize now has two seasons in the market: 58% arrives from October to January from the kharif crop and 32% from March to August from rabi and summer. Moong arrives from June to September (56%), the summer crop first and the kharif crop after. Onion, potato and tomato, grown in several seasons and regions, arrive at 6–11% of the year’s total every month.
That concentration is why harvest-month prices are so often the year’s lowest, and why procurement, storage and warehouse-receipt credit matter most in exactly those weeks: in April for wheat, in October and November for paddy, soybean and kharif maize.
What reaches the mandi is not keeping pace with what is grown
| Crop | Mandi arrivals, million t, 2019 → 2024 | Production, million t, 2018-19 → 2024-25 | Arrivals as share of production, 2024 |
|---|---|---|---|
| Maize | 4.8 → 9.4 (+97%) | 27.7 → 43.4 (+57%) | 22% |
| Paddy / rice | 30.5 → 36.3 (+19%) | 116.5 → 150.2 (+29%) | not comparable |
| Wheat | 27.0 → 26.8 (−1%) | 103.6 → 117.9 (+14%) | 23% |
| Soybean | 6.4 → 6.6 (+3%) | 13.3 → 15.3 (+15%) | 43% |
| Groundnut | 1.5 → 2.4 (+62%) | 6.7 → 11.9 (+78%) | 20% |
| Mustard | 3.1 → 2.9 (−5%) | 9.3 → 12.7 (+37%) | 23% |
| Gram | 3.2 → 1.9 (−41%) | 9.9 → 11.1 (+12%) | 17% |
| Moong | 0.7 → 0.7 (+3%) | 2.5 → 4.2 (+73%) | 17% |
| Tur | 1.0 → 0.9 (−8%) | 3.3 → 3.6 (+9%) | 26% |
| Cotton | 5.4 → 3.8 (−30%) | 28.0 → 29.7 (+6%) | not comparable |
Comparing arrivals in calendar 2019 and 2024 with production in the crop years that feed them, before Agmarknet widened its market coverage in November 2025, the share of each crop that passes through regulated mandis has moved in different directions. Maize arrivals rose +97%, faster than production (+57%), as ethanol distilleries and feed mills bought more through markets. Groundnut arrivals also rose. But gram arrivals fell −41% while production rose +12%; mustard arrivals fell slightly while production rose +37%; moong production rose +73% with arrivals almost flat; and wheat arrivals were unchanged while wheat production rose +14%. The data cannot say why. Likely reasons include government procurement at centres that do not report as mandis, direct sales to processors and traders outside the APMC system, and markets that report irregularly; the repository’s notes list these caveats.
What doesn’t follow from any of this
A rising rabi share does not mean kharif matters less: kharif still grows almost all of India’s rice outside the boro crop, most of its oilseeds and pulses, and all of its cotton and soybean, and a failed monsoon still does the most damage. The yield gap between seasons reflects irrigation and crop mix as much as season; rabi wheat and kharif bajra are not the same crop on the same land. Summer data cover only four years, the latest year is an advance estimate, and arrivals measure what markets report, not what farmers sell. Arrival calendars use 2019–2024 to avoid the November 2025 change in Agmarknet coverage, and annual comparisons mix calendar and crop years.
What would fix this
- The Department of Agriculture should publish summer-season estimates back to at least 2010-11 by separating them from the existing rabi series, so that the growth of zaid crops can be tracked over more than four years.
- Agmarknet should publish a coverage index, the number of reporting markets by state and month, alongside its arrivals data, so that a rise in arrivals can be separated from a rise in the number of markets reporting.
- Procurement agencies (FCI, NAFED, NCCF and state agencies) should report their purchases by state and month in the same format as mandi arrivals, so that the share of a crop sold through any formal channel can be measured, not only the share sold in regulated mandis.
- State agriculture departments should publish groundwater use for summer rice, maize and moong, the three crops driving zaid growth, before extending input subsidies or support prices to the season.
These are proposals from this blog, not recommendations any agency has adopted.
Sources and caveats
Area, production and yield: Department of Agriculture and Farmers Welfare, Unified Portal for Agricultural Statistics (UPAG), all-India estimates by season and crop, 1966-67 to 2025-26 (2025-26: third advance estimates), and state estimates for 2025-26 (summer and rabi leaders). Mandi arrivals: Agmarknet (Directorate of Marketing and Inspection), state-wise monthly arrivals, January 2019 to October 2026. Both are as extracted in the open repository herrrickshaw/india-mandi-arrivals (files data/upag/apy_all_india.csv.gz, data/upag/apy_state.csv.gz and data/agmarknet/state_monthly.csv.gz), which documents the endpoints, the November 2025 coverage change and other caveats. Season definitions follow the ministry’s usage; summer (zaid) is reported separately from 2022-23 and is added to rabi here for comparisons over time. Cotton production is in bales and its mandi arrivals are seed cotton, so the two are not comparable; paddy arrivals are unmilled grain, compared here with milled rice production only in direction. Three-year averages, shares, volatility and arrival calendars are this blog’s arithmetic. Research notes: out/research/seasons/ in this blog’s archive.
About this article: Researched, written and edited by Umashankar Triplicane Dwarakanathan, with AI research assistance; every figure is meant to trace to the primary source cited. See the Editorial Policy for how sourcing, AI use and corrections work.
