To explore spatial variability of agricultural soil moisture and salinity on microscales under mulch drip irrigation, we conducted filed experiments and employed classical statistics and geostatistics to investigate the spatial variability of soil moisture and salt content in cotton field on three microscales (0.25 m×0.25 m, 1 m×1 m, and 4 m×4 m) and in different soil depths and determine the reasonable sampling number. The results indicated that the variability of soil moisture was moderate and relatively weak on three microscales, which was enhanced as scale increased and enhanced first and then weakened as depth increased. The variability of soil salinity was moderate and relatively strong, which was enhanced as scale increased and enhanced first and then weakened as depth increased. On three microscales and in different depths, most semi-variograms of soil moisture and salinity could be simulated by Gaussian models with high accuracy and the reasonable sampling number should be 367. The results could provide theoretical guidance for developing monitoring scheme and control measures for soil moisture and salinity under mulch drip irrigation.