A semiempirical model is developed to represent the seasonal profile of the normalized difference vegetation index (NDVI) over temperate agricultural regions observed by NOAA/AVHRR local area coverage measurements where, due to 1.1 km resolution, pixels are mixtures of agricultural crop fields. However, crops with similar phenology behave radiatively as optically homogeneous canopies. In such cases, a logistic function using five parameters describes the annual time profile of the NDVI. Two of the five parameters are the ascending and descending inflection points, which convey important information about the development of the crop. Over a region like Beauce (approximately 10,000 km2), the relative area and phenology of the crops grown results in two time profiles, one for winter crops and one for summer crops. Consequently, two logistic functions were combined to describe the variations of the NDVI derived from NOAA /AVHRR data. The equations are fitted using a nonlinear least squares procedure against data sets that included hand-held radiometer measurements in fields of the individual crops, and two kinds of AVHRR data: actual data and synthetic data contaminated by artificial noise. The use of synthetic data indicates that the inflection points are retrieved with good accuracy, even with noise. When the inversion scheme is applied to actual AVHRR data, it appears that a priori knowledge about the asymptotic values reached by the NDVI of the various crops is required in order to correctly estimate the dates of the inflection points, and then to obtain some useful information about the development of the crops.