When I attempt a sigmoidal curve fit, Prism mistakenly tells me that I should enter the logarithm of each concentration, rather than the concentration itself, in the X column.

The convention for this type of graph is to plot measured quantity vs. log of the molar concentration. Prism checks your X values to see if they are within an expected range. If they are not, there is the possibility that you haven't log transformed the data first and that use of that model is invalid. Prism displays this--or a similar--advisory message:

However, using concentration units other than molar, such as picomolar or ng/mL, may cause the log X values to fall outside the range that Prism expects. If this is the case, you can just click "OK"; Prism will complete the analysis. Prism can't know whether the values are logarithms, so can sometimes display the warning even when the values are already logs.