We analysed land cover changes and associated farm system changes in the Mar Chiquita basin, a highly productive agricultural area in the Argentinian Pampa region. Changes in land cover between 1999 and 2013 were analysed to provide better insight into the differences obtained by analysing land use change at pixel level and at the level of farm units. Land cover change was characterized using multitemporal satellite images, while farm systems were subsequently identified based on cadastral land holdings. Results show that the area covered with grassland has decreased, while the area covered with soybean, winter cereals-soybean double cropping, other crops, and horticulture has increased; the first two with 166% and 158% respectively. Over the same period, the number of livestock farms has decreased by 41%, while the number of cropland farms, horticultural farms and mixed farms have increased. We studied the location of land cover, land cover changes, farm systems and farm systems changes by relating spatial patterns of land use (change) to a set of potential location factors. The location of different land cover types was explained fairly well by this set of socio-economic and biophysical location factors, while the allocation of land cover changes was explained poorly. Farm system locations, on the other hand, were explained with very high accuracy, and farm system changes with a good accuracy. We conclude that farms are a more appropriate unit for land use change analysis than pixels in areas dominated by agricultural use and featuring crop rotations. Analysis at the farm level reduces the noise caused by crop rotations. As farms are the units at which land use decisions are implemented insight in the changes at this level has the potential to better inform land management practices, land use planning and policies.