We used budgets of absorption (a), scattering (b), and backscattering (bb) for particles and chromophoric dissolved organic matter (CDOM) to investigate sources of seasonal variations in apparent optical properties (AOPs) of New England continental shelf surface waters. Spectral a, b, and bb budgets for particles were estimated from flow cytometric measurements of eukaryotic pico/nanophytoplankton, Synechococcus, heterotrophic prokaryotes, detritus, and minerals; AOPs were modeled with Hydrolight radiative transfer software. For late summer and spring, our modeled values of the diffuse attenuation coefficient (Kd) and remote sensing reflectance (Rrs) were on average within 15% and 9%, respectively, of independent measurements. This close agreement allowed us to examine how different seawater constituents contributed to AOP variability. Higher values of Kd in the spring, compared to summer, were due to higher absorption by eukaryotic phytoplankton (aeuk) and CDOM (aCDOM), which coincided with higher nutrient levels and less stratified conditions than in the summer. Differences in the spectral shape of Rrs between the seasons were caused by a combination of differences in aeuk, aCDOM, and bb from non-phytoplankton particles (minerals and detritus combined). For non-phytoplankton bb the major seasonal difference was a higher inverse wavelength dependence in the summer due to the effects of small organic detritus. We applied two semianalytical ocean color models to our data, in order to evaluate whether the assumptions and parameterizations inherent in these models are applicable for New England shelf waters. We show how differences between observed and modeled chlorophyll a specific phytoplankton absorption, aCDOM, and non-phytoplankton bb cause errors in chlorophyll a concentration and IOPs retrieved from reflectance inversion models.