The propagation of downwelling irradiance at wavelength λ from surface to a depth (z) in the ocean is governed by the diffuse attenuation coefficient, $bar{K}_{d}$(λ). There are two standard methods for the derivation of $bar{K}_{d}$(λ) in remote sensing, which both are based on empirical relationships involving the blue-to-green ratio of ocean color. Recently, a semianalytical method to derive $bar{K}_{d}$(λ) from reflectance has also been developed. In this study, using $bar{K}_{d}$(490) and $bar{K}_{d}$(443) as examples, we compare the $bar{K}_{d}$(λ) values derived from the three methods using data collected in three different regions that cover oceanic and coastal waters, with $bar{K}_{d}$(490) ranging from ~0.04 to 4.0 m-1. The derived values are compared with the data calculated from in situ measurements of the vertical profiles of downwelling irradiance. The comparisons show that the two standard methods produced satisfactory estimates of $bar{K}_{d}$(λ) in oceanic waters where attenuation is relatively low but resulted in significant errors in coastal waters. The newly developed semianalytical method appears to have no such limitation as it performed well for both oceanic and coastal waters. For all data in this study the average of absolute percentage difference between the in situ measured and the semianalytically derived $bar{K}_{d}$ is ~14% for λ = 490 nm and ~11% for λ = 443 nm.