Radar backscatter depends mainly on the characteristics of observed covers and on the configuration of the sensor. In agricultural areas, as crops grow, their characteristics vary, and their backscattering behavior changes. The main objective of this study is to perform a multitemporal analysis of the variations in backscatter coefficients and ratios at different polarizations due to the growth stage of each crop and, in turn, to establish the optimal dates for accurate crop separation and classification. With this aim, five RADARSAT-2 scenes were acquired over the Pamplona basin (North of Spain) between March and June 2010, covering the major part of the growing season of agricultural crops in this area. The results obtained illustrate the importance of May and June acquisitions, particularly in VV and HV polarizations, for obtaining adequate crop separabilities and accurate classification results. Cereals showed a particular behavior with a stronger attenuation in VV polarization than in HH, due to the vertical orientation of stems, during the vegetative growth stages. In early June, barley had a peak in HV polarization, revealing an earlier heading compared to wheat and oats. Rapeseed and peas had a strong volume scattering contribution due to their heterogeneous canopies, and sunflower could be discriminated because of its different agricultural calendar. Crops were classified with an accuracy of 80% using just three RADARSAT-2 scenes acquired between May and June.