The　heterogeneous　composition　of　people　living　in　various　areas　creates　differences　in　the　urban　evacuation　demand　after　an　earthquake.　In　this　study,　four　factors　that　can　influence　the　evacuation　demand　are　discussed:　types　of　people,　seismic　intensity,　evacuation　pattern,　and　road　capacity.　Based　on　the　formation　mechanism　of　the　evacuation　demand,　a　mathematical　model　for　evacuation　demand　evolution　is　established　through　the　susceptible-infective　(SI)　theory　to　address　these　factors　and　their　effects.　Considering　the　dynamic　feedbacks　during　the　evacuation　process　and　the　spatiotemporal　characteristics　associated　with　the　types　of　people,　a　microcosmic　dynamic　feedback　model　for　evacuation　demand　prediction　is　proposed　coupled　with　the　system　dynamics　(SD)　theory.　Through　comparative　simulation　of　evacuation　units　experiencing　the　earthquake　at　different　times　and　variations　in　land-use　functions,　the　evolution　characteristics　of　evacuation　were　simulated　more　intuitively.　The　dynamic　prediction　and　evaluation　model　based　on　SI　and　SD　in　this　paper　can　provide　a　new　method　for　evacuation　demand　prediction.　The　analysis　based　on　the　simulation　is　of　guiding　significance　for　the　evaluation　of　earthquake　disaster　prevention　losses　and　the　implementation　of　disaster　prevention　planning.　It　can　also　provide　a　reference　for　the　prediction　and　decision-making　of　similar　problems.