When entering data for a log(dose) response curve, should I use common (base 10) or natural (ln) logarithms?

 Use common base10 logarithms (review the difference) for two reasons:

• After fitting the log of the EC50, Prism does the antilogarithm transform to report the EC50 and its logarithm. Prism assumes that you used base 10 logarithms, so does the reverse transform by taking ten to the power. 
• The best-fit value of the Hill Slope would be different if you used natural logarithms, and so the value cannot be compared to other Hill slopes.

This is just convention, and one could certainly write equations and interpret Hill slopes using natural logs. But, at least for pharmacologists, use of natural logarithms is... unnatural. Base 10 logs are used exclusively.