Current ocean color algorithms based on remote-sensing reflectance spectra, Rrs(λ), overestimate chlorophyll a concentrations, Chl, and particulate backscattering coefficients, bbp(λ), in optically shallow oceanic waters due to increased bottom reflectance. Since such regions often contain important ecological resources and are heavily influenced by human populations, accurate estimates of Chl and bbp(λ) are essential for monitoring algal blooms (e.g. red tides), detecting sediment resuspension events and quantifying primary productivity. In this study, a large synthetic data set of 500 Rrs(λ) spectra is developed to examine limitations of ocean color algorithms for optically shallow waters and to develop alternative algorithms that can be applied to satellite (e.g. SeaWiFS and MODIS) and aircraft ocean color sensor data. Rrs(λ) spectra are simulated using a semi-analytic model for optically shallow waters. The model is parameterized with sand bottom albedo spectra, ρ(λ), using a wide range of chlorophyll a concentrations (0.03-30 mg m-3), bottom depths (2-50m) and bottom albedos (ρ(550)=0.01-0.30) to provide a robust data set that accurately represents and complements shipboard Rrs(λ) data from the Gulf of Mexico and Bahamian waters. The accuracy of a remotely-based technique developed recently from shipboard Rrs(λ) data is tested on the synthetic data for identifying waters with bottom reflectance contributions at Rrs(555) greater than 25%. Limitations and improvements regarding this method are discussed.