Vibrational sum-frequency generation (VSFG) has become a widely used technique for studying molecular orientation and intermolecular structure at interfaces. Due to the interfacial and vibrational selectivity of this technique, it can be used to probe molecular monolayers that are buried within bulk materials. Here, we present a critical review and examination of the assumptions that are generally used in the interpretation of VSFG spectra. This review focuses on three different aspects of VSFG spectroscopy. First, we examine the effect of dynamics (both reorientation and energy transfer) on the interpretation of VSFG spectra. Second, we consider whether (and under what circumstances) VSFG spectra cannot be interpreted solely in terms of the spectral properties of individual molecules. Third, we consider whether VSFG spectra obtained in different modes (frequency-domain, time-domain and broadband) should provide identical information. We conclude with a discussion of the challenges and opportunities provided by the revised viewpoint of VSFG that we present.