Over the past decade, many investigators have published techniques for deriving phenological parameters, especially the start of the growing season (SOS), from time-series satellite imagery. The principal satellite sensor for these studies is the advanced very high resolution radiometer (AVHRR). This study investigates the characteristics of four of the primary methods for identifying SOS; maximum curvature, delayed moving average, time of greatest NDVI increase and time of half-maximum NDVI. Comparisons between the methods illustrate biases for earlier or later SOS in the various methods and a relatively low statistical relationship among all four methods (r²=.10 to .35). Our analysis indicates that each of the methods is detecting fundamentally different stages of the start of season. These stages include snow melt, initial growth of herbaceous plants and understory, and first leaf; each of which is important in determining differences in land/atmosphere interactions. Users of these types of datasets need to be aware of the different environmental conditions being measured by the various approaches and use careful judgment in selecting the proper method.