Abstract In the micrometeorology community, it is well known that the turbulent fluxes measured with eddy covariance (EC) systems do not usually equal the available energy. Hence, qualitative knowledge of the impact of different vegetation types, and climatic variables on this ‘nonclosure’ is essential. This study analyzed a unique database of EC flux measurements covering 8 growing seasons of 3 crops (maize, wheat, and rapeseed) cultivated over two close agricultural sites (FR-Lam and FR-Aur) in southwestern France. For data analysis, some dry and wet cropping seasons of the same crop type were selected; then, their phenological stages were identified to investigate their effect on the energy balance closure (EBC), and flux partitioning. The results showed that the systematic effect of each site on the EBC was stronger than the influence of crop type and stage, as EBC was generally higher at FR-Aur (82%) than at FR-Lam (67%), even for the same crop type. The assessed effect of rainfall, and phenological stages on energy partitioning revealed that during the wet seasons, over 42% of the net radiation (Rn) was accounted for by the latent heat flux (LE), which was 9% higher than the recorded LE in the dry year during the active vegetation period. Similarly, the ground heat flux (G) was observed to be very sensitive to vegetation; G accounted for 30% of Rn when vegetation was low, whereas at the peak of vegetation, it fell below 16% due to canopy shading. Closure was also assessed under various atmospheric stability conditions and wind sectors, and it was observed to be higher under unstable conditions, and in prevailing wind directions. Analysis of the sensible heat advection (AH) revealed that AH accounts for more than half of the imbalance at both sites.