The expansion of sugarcane crop to regions with lower water supply in Brazil has increased the importance of correct estimation of crop water requirements. Currently, the irrigation management is generally done using the crop coefficient (Kc) based on the FAO 56 bulletin. Kc is used to determine the potential water demand of the crop for a given period of time and is considered constant for each crop stage. However, some recent studies have shown that Kc can be significantly variable under different evapotranspiration (ETo) rates. This paper aimed to analyse sugarcane water consumption at different scales: plant (sap flow measurements by energy balance method); canopy (Bowen ratio energy balance method); and plant–atmosphere coupling (infrared gas analyser) to reduce the uncertainties on the irrigation practices. Measurements were taken at two experimental sites, where a modern Brazilian cultivar CTC 12 was grown under drip irrigation and an old main Brazilian cultivar (RB867515) was grown under sprinkler irrigation by a central pivot. The mean crop evapotranspiration (ETc) values by the Bowen ratio energy balance method were 2.92 and 3.68 mm d−1 for RB867515 and CTC 12, respectively, resulting in a mean Kc of 0.99 at the full vegetative growth stage. Kc values were dependent on ETo and varied between 0.2 and 1.7 for both cultivars. This occurred in a crop coupled to the atmosphere (Ω = 0.37) and was the same found in other coupled crops such as coffee and citrus. In conclusion, the sugarcane Kc for southeast Brazil presented temporal variability due to coupling conditions according to reference evapotranspiration, and this should be considered in irrigation management.