A significant cause of greenhouse gas emissions in agriculture is claimed to be that produced by burning fossil fuel during various farming operations. Few long-term results exist, however, for Australian cotton farming systems. This paper presents emissions estimated from fuel and electricity use in an experiment established in 1985 at the Australian Cotton Research Institute, near Narrabri, NW New South Wales, Australia. Treatments were continuous cotton sown after either conventional (disc- and chisel ploughing, followed by ridging) or on “permanent beds” (slashing of cotton plants after harvest, followed by root cutting and bed renovation with a disc-hiller), and cotton-wheat rotation sown on “permanent beds”. Until 1999 wheat stubble was incorporated and since 2000 was retained as in-situ mulch before sowing cotton. Conventional cotton varieties were sown until 1999, Round-up Ready varieties until the 2005, and “Bollgard-Roundup Ready-Flex” varieties thereafter The experiment was laid out as a randomised block design and replicated 4 times in plots 190 m long and 36 rows wide. Cotton crops received 160-180 kg N/ha during August of each year as anhydrous ammonia before sowing cotton until the 2008-09 season, and as urea after sowing cotton thereafter. Urea was applied to wheat before sowing at a rate of 20 kg N/ha, and 60 kg N/ha subsequently during later July or early August. Cotton and rotation crops were irrigated at an average rate of 100 mm subject to water availability, rainfall and soil water content. Greenhouse gas emissions (as carbon dioxide equivalents, CO₂-e) associated with farming operations, herbicide and fuel production and transport were estimated from available sources in the literature by relating them to diesel and electricity consumed from June 1998 to July 2010. Key findings were:

• Groundwater pumping was a major contributor to greenhouse gas emissions. Emissions produced during irrigation with river water were relatively small whereas in comparison, those produced by pumping groundwater were about 6 times higher.
• Reducing tillage reduced in-field emissions whereas continuous cotton increased them. Reduction of in-field CO₂-e production was of the order of 28% on a per hectare basis, 32% per unit of cotton lint produced and 62% per unit of energy in seed products (wheat and cotton) when management practices such as permanent beds and wheat rotation crops were used in comparison with sowing cotton every year after conventional tillage.
• Least in-field emissions were produced when wheat was part of the rotation. The lower emissions were due mainly to the inclusion of a stubble-mulched fallow, that reduced the number of farming operations while concurrently facilitating the harvesting of rain water, thereby reducing the number of irrigations. 
• Partial life cycle analysis showed that most CO₂-e was produced by in-field activities and during production of N fertiliser used in the field.

In summary, water conservation practices such as retaining stubble as in-situ mulch, reducing tillage intensity and optimising irrigation were most effective in reducing greenhouse gas emissions from a long-term experiment in an Australian Vertisol. Past research has also shown that these are the very same practices which can improve soil health, cotton yield and fibre quality. The above analysis does not account for losses from the soil in the form of N₂O from fertiliser and decomposing crop residues, and CO₂ from mineralisation of soil organic matter and crop residues.