Sea surface temperature retrievals using the Advanced Very High Resolution Radiometer are highly sensitive to cloud cover and coarse mode aerosol particles such as dust. Operationally, techniques are used to flag contaminated retrievals; however, these techniques are less precise in removing dust-contaminated values. A commonly stated metric of quality for SST daytime retrievals is 0.5°C; thus dust contents that produce errors greater than this value should be of concern. Here we report on significant correlation between potential SST error and observed aerosol optical depths (AOD) that was found in the tropical region dominated by Saharan dust. Utilizing a radiative transfer model with variable dust contents and typical vertical distributions, errors greater than the desired 0.5°C accuracy are observed for dust AODs as low as 0.05. Errors of over 1°C occur with 0.25 AOD. Analysis of the AERONET data from Cape Verde, which includes the Saharan Air Layer off the west coast of Africa, reveals that 90% of the days during the boreal summer are found to have AOD amounts that correspond to error greater than 0.5°C. We found that a correction accurate within 0.25°C requires a mean accuracy of 0.1 AOD and proper vertical placement of the dust layer within 250m. While empirical SST retrievals often have some measure of climatological dust contamination built into them, this work shows that typical variability in dust loadings is a non-trivial error source against SST retrieval goals. •We model and observe the sensitivity of airborne dust particles on SST retrievals.•A clear increasing negative bias in SST occurs with increasing AOD.•Seasonality of dust can lead to season and regional biases in SST.•A dust AOD of 0.1 roughly corresponds to an SST error of 0.5°C.•Dust corrections must include the vertical location and temperature of dust layer.