Comet C/2013 A1 (Siding Spring) will have a close encounter with Mars on October 19, 2014. We model the dynamical evolution of dust grains from the time of their ejection from the comet nucleus to the Mars close encounter, and determine the flux at Mars. Constraints on the ejection velocity from Hubble Space Telescope observations indicate that the bulk of the grains will likely miss Mars, although it is possible that a few-percent of grains with higher velocities will reach Mars, peaking approximately 90--100 minutes after the close approach of the nucleus, and consisting mostly of millimeter-radius grains ejected from the comet nucleus at a heliocentric distance of approximately 9~AU or larger. At higher velocities, younger grains from sub-millimeter to several millimeter can reach Mars too, although an even smaller fraction of grains is expected have these velocities, with negligible effect on the peak timing. Using NEOWISE observations of the comet, we can estimate that the maximum fluence will be of the order of $10^{-7}$ grains/m$^2$. We include a detailed analysis of how the expected fluence depends on the grain density, ejection velocity, and size-frequency distribution, to account for current model uncertainties and in preparation of possible refined model values in the near future.