Geologic features resembling terrestrial water-carved gullies imply that li
quid water has flowed recently on the surface of Mars and challenge our vie
ws of the present-day low-temperature environment. We evaluate two possible
mechanisms for the formation of liquid water under environmental condition
s that we expect to have existed on Mars in its recent past. First, we exam
ine the stability of ground ice in the permafrost and the potential for mel
ting near-surface ground ice (in the top few meters of soil) by solar heati
ng and subsurface conduction. Second, we examine the potential for melting
and refreezing of ice at shallow depths due to geothermal heating. We find
that near-surface ground ice does not reach the melting point of water unde
r a range of conditions of soil thermophysical properties, latitudes, obliq
uities, and surface slopes. The atmosphere remains too dry for the ground i
ce to melt, even at high obliquity; instead, ice sublimates before reaching
melting temperatures. However, the presence of salts in concentrations of
15-40% can adequately lower the melting point to allow melting to occur. We
also find that a combination of a global average geothermal heat flux and
a thick, low-conductivity, unconsolidated regolith raises the depth of the
melting isotherm to less than a few hundred meters from the surface. Orbita
lly induced oscillations in the mean annual surface temperature can cause f
reezing cycles in a confined aquifer at this depth. Freezing pressures gene
rated are adequate to fracture ice-cemented ground and allow water to escap
e to the surface, similar to the formation and evolution of terrestrial pin
gos in shallow permafrost. Both mechanisms are possible; however, the geoth
ermal mechanism is consistent with the observations of the distribution of
gullies, while the salty near-surface ground ice mechanism is not. Further
observational tests that can be performed with existing and future spacecra
ft are suggested.