Soil erosion is a major environmental problem that can lead to loss of fertility and degradation of agricultural fields. In order to develop efficient strategies to mitigate the impact of precipitation and reduce the erosion rate, a process-based understanding of the mechanisms that govern sediment transport and delivery is necessary. Soil state and physical properties prior to a precipitation event can affect significantly the erosion rate. Among the most important soil variables are moisture content, compaction and infiltration capacity. Additionally, the presence of stones on the topsoil surface retards the overland flow discharge, reduces runoff generation as well as the sediment delivery and prevents the development of a surface seal, which in turn maintains the infiltration rate. The aim of this study was to examine in detail the effect of surface stones, soil compaction and sealing for a sequence of rainfall events on soil erosion. Experiments were conducted using the EPFL erosion flume, which was divided into two identical flumes (one with stone and one without). The experiment involved four rainfall events with the precipitation rates: 28, 74, 74 and 28 mm h-1. After each 2-h event, the soil was allowed to air dry for 22 h. The total sediment concentration, the concentration of seven sediment size classes and the flow discharge were measured during each event at the outlet of each flume. Experimental results were analyzed using the Hairsine and Rose (H-R) soil erosion model. Results showed that (i) within each precipitation event, the proportion of each size class for the bare/stone-covered flume pairs at steady state were similar, whereas the initial response differed significantly; (ii) in all cases the effluent was enriched in finer particles relative to the original soil; and (iii) the effluent sediment composition was different from that of the original soil, and there was no clear trend towards the parent soil sediment size composition with time. The H-R model was able to reproduce well the events with high precipitation rate (events 2 and 3) with the same parameter set, while the match was less satisfactory for the low precipitation events. A possible explanation for this is that the initial soil compaction/sealing/development of the deposited layer combined to yield a surface that eroded similarly for identical rainfall conditions. Changes in the precipitation rate modifies the soil surface (the deposited layer in particular) and thus the erosion rates. Model application further suggested that over the course of the rainfall events, the contribution of the original soil to the eroded sediment decreased gradually, while that of the deposited layer increased.