Above-water reflectance and surface chlorophyll a concentrations (Chl a) were measured in the Gulf of Lions, northwestern Mediterranean Sea in 2000 and 2001 in order to test Chl a inversion algorithms. Surface waters were separated in Case 2 waters in the Rhône River plume and proximal Region of Freshwater Influence (ROFI) stations, and Case 1 waters at all the other stations. Case 2 waters were characterized by R443/R555 < R443/R510 < R490/R555 < R490/R510 < 1. In the first part, we compared the concurrent reflectance measurements made with a scanning polarization radiometer (SIMBAD) and a hyperspectral Ocean Optics radiometer. The comparison of the remote-sensing reflectance (Rrs) values at SIMBAD wavelengths shows excellent agreement for Rrs values higher than 0.01 sr-1. Between the two instruments, reflectance ratios, commonly used in Chl a algorithms, show differences smaller than 2% in the Case 2 waters, and smaller than 20% in the Case 1 waters. In the second part, concurrent measurements of Chl a and of hyperspectral reflectance from 6 cruises were used to analyze the statistical performance of global (OC2, OC4) and regional regression algorithms using mainly SeaWiFS bands. The algorithms were tested first over the entire domain, then separately over the Case 1 and Case 2 waters. Chl a algorithms using band ratios such as the one presented in Bricaud et al. (2002) are suitable for the Case 1 waters. However, taking into account the large dispersion of Chl a for very close reflectance ratios in the Case 2 waters, single band ratios are not suitable for deriving Chl a. The use of a 4-wavelength parameter such as Xc, defined by Tassan (1994), leads to better results in the plume and proximal Rhône ROFI.