Soil erosion by water is one of the most severe soil degradation processes. Splash erosion is the initial stage of soil erosion by water, resulting from the destructive force of rain drops acting on soil surface aggregates. Apart from rainfall properties, constant soil physical properties (texture and soil organic matter) are crucial in understanding the splash erosion. However, there is lack of information about the effect of variable soil properties such as soil initial water content and surface condition (seal formation) on splash erosion. The objective of the present study was to determine how initial water content and surface condition affected soil splash erosion under simulated rainfall. The changes in soil surface condition were characterized by hydraulic variability (saturated hydraulic conductivity) due to surface seal formation. Slit loam and loamy sand soil textures were used in the experiment. The soil samples were collected from the top layer; air dried, and filled into modified Morgan splash cups for splash erosion measurements. Rainfall was created in the laboratory using two types of rainfall simulators covering intensity range from 28 to 54 mm h−1 and from 35 to 81 mm h−1. The soil samples were exposed to three consecutive rainfall simulations with different time intervals between simulations and different initial water content and surface conditions (air-dried, wet-sealed, and dry-crusted). Wet-sealed soil samples had up to 70% lower splash erosion rate compared to air-dried samples, due to surface ponding followed by seal formation. A significant decrease in soil saturated hydraulic conductivity indicated the formation of surface seal for silt loam soils. A non-significant decrease in saturated hydraulic conductivity for loamy sand soil was attributed to earlier formation of stable seals. Two different rainfall simulators produced different amount of splash erosion rates; however, the splash erosion development for increasing rainfall intensity was almost equal considering same initial surface condition. These results provide insight into dynamic changes of individual soil parameters affected by rainfall, and could find wider application for more complex soil erosion prediction models.