Since GM crops arrived in 1996, US pesticide use rose modestly overall — almost entirely because of a near-10x jump in glyphosate — while corn and soybean yields kept climbing at essentially the same rate they were climbing before GM traits existed

Summary

Pesticide use: modestly up in total pounds, but only because of one chemical. Summing USGS's county-level agricultural pesticide-use estimates nationally, total pesticide active ingredient applied to US farms rose about 16% (low estimate) to 4% (high estimate) between 1996 and 2018. Almost the entire net increase — and then some — is explained by glyphosate, the herbicide paired with Roundup Ready genetically modified (GM) crops: glyphosate use grew roughly 9-fold (12.3 to 122.0 million kg), while every other pesticide combined actually fell about 10%. So the honest one-line answer is: total pesticide volume is up modestly, insecticide use is down substantially (mainly from Bt corn/cotton), and herbicide use is up substantially and specifically because of glyphosate — a pattern every major independent study (USGS, USDA ERS, Benbrook, National Academies) agrees on, even though sources disagree sharply on how to weigh it.

Yields: kept rising, but not because GM traits changed the underlying trend. US corn yield rose from 127.1 to 179.3 bushels/acre (1996-2024), a 41% gain; soybean yield rose from 37.6 to 50.7 bu/acre, a 35% gain. But the National Academies of Sciences' 2016 review and USDA/university agronomists both conclude the rate of yield improvement did not accelerate when GM crops arrived — corn's 1996-2024 gain averages 1.86 bu/acre/year, essentially identical to the ~1.9 bu/acre/year rate that has held since hybrid corn's second productivity "miracle" began in the mid-1950s. GM traits mainly protected existing yield potential from weeds, insects, and (for herbicide-tolerant varieties) reduced weed competition, rather than raising the genetic yield ceiling itself.

What the pesticide data actually shows

Querying USGS's NAWQA Pesticide National Synthesis Project county-level estimates (ag.pesticide_use_by_county, active-ingredient kilograms, 1992-2018 with the low/high estimate methods USGS itself publishes) and summing nationally:

This decomposition — that the WHOLE net national increase, and more, traces to glyphosate — is a figure this analysis computed directly from the primary USGS series; it sharpens (rather than merely repeats) the qualitative story in the literature. It matches the mechanism described by USGS scientists Coupe and Capel (2016, Pest Management Science): corn's per-acre herbicide rate actually fell after GM introduction, soybean herbicide use rose, and cotton was roughly flat — but glyphosate specifically displaced many older, often more toxic, herbicide products across all three crops. It is also consistent with Charles Benbrook's widely cited estimate (Benbrook 2012, Environmental Sciences Europe) that herbicide-tolerant technology added about 239 million kg of herbicide 1996-2011 while Bt insect-resistant traits cut insecticide use by about 56 million kg over the same span — net pesticide use up roughly 7% by his count, with the increase concentrated in herbicides and driven by the spread of glyphosate-resistant "superweeds," which by his account was pushing up herbicide volume by roughly 25% a year in the worst-affected fields. The Union of Concerned Scientists and USDA's own Economic Research Service reach the same qualitative split (insecticides down substantially, herbicides up, driven by resistant weeds), though estimates of the exact magnitude vary by study and vintage — Benbrook's numbers, for instance, run through 2011 or 2016 depending on the paper, while the USGS series used here runs through 2018, the last complete year in the ag.pesticide_use_by_county table (2019 is only partially loaded and was excluded; USGS also discontinued seed-treatment pesticide estimates starting in 2015, which are not counted in any of these figures for any year).

What the yield data actually shows

USDA NASS's own official annual yield series (fetched directly from nass.usda.gov, cross-checked against the 2012 drought value of 123.1 bu/acre and the record 2025 value, both well-known landmark figures) shows corn yield climbing from 127.1 bu/acre in 1996 to 179.3 in 2024 (+41.1%) and a preliminary 186.5 in 2025 (+46.7%). Soybean climbed from 37.6 to 50.7 bu/acre over the same span (+34.8%), 53.0 in the 2025 preliminary estimate. The key finding from the National Academies of Sciences, Engineering, and Medicine's 2016 report "Genetically Engineered Crops: Experiences and Prospects" — the most comprehensive independent review of GE-crop evidence to date — is that the committee found no evidence GE crops changed the rate of yield increase: yields rose at a steady pace spanning both the pre-biotech and biotech eras, which the committee attributed mainly to continued advances in conventional breeding and agronomy rather than the transgenic traits themselves. This analysis independently reproduces that pattern: the corn series' 1996-2024 average annual gain (1.86 bu/acre/year) matches the ~1.9 bu/acre/year rate that Purdue agronomist Bob Nielsen's historical analysis (citing the same NASS series back to 1866) documents as the trend since hybrid corn's mid-1950s productivity shift — a rate that, per Nielsen, "has continued... ever since" with "little to no evidence" that biotech traits introduced starting in the mid-1990s pushed it any higher. The Academies' report did note one nuance: herbicide-tolerant (glyphosate-resistant) varieties specifically helped growers control weeds better, which can raise realized yield by reducing weed competition — but this shows up as protecting yield potential in a given field/year, not as raising the crop's underlying genetic ceiling or its long-run trend rate.

Reconciling the two halves of the answer

Put together, the honest answer to "more or less pesticide, and what happened to yields" is not a single clean number in either direction: total pesticide poundage is up modestly, but that increase is entirely a glyphosate story riding on top of a roughly 10% decline in every other pesticide (insecticides fell the most, mainly from Bt corn and cotton reducing insect-control sprays) and yields kept rising throughout, but on the same statistical trend line they were already on before 1996 — so GM adoption coincided with continued yield growth without being shown, in the National Academies' review of the evidence, to have caused an acceleration in that growth. The glyphosate-resistant-weed dynamic that literature attributes to the herbicide increase (Benbrook 2012; Coupe & Capel 2016) is also the main reason per-acre herbicide rates have crept up again in the 2010s after an initial post-1996 decline, since more glyphosate per application became necessary as weeds evolved resistance.