ABSTRACTRepeatability of seismic data plays a crucial role in time‐lapse seismic analysis. There are several factors that can decrease the repeatability, such as positioning errors, varying tide, source variations, velocity changes in the water layer (marine data) and undesired effects of various processing steps. In this work, the complexity of overburden structure, as an inherent parameter that can affect the repeatability, is studied. A multi‐azimuth three‐dimensional vertical‐seismic‐profiling data set with 10 000 shots is used to study the relationship between overburden structure and repeatability of seismic data. In most repeatability studies, two data sets are compared, but here a single data set has been used because a significant proportion of the 10 000 shots are so close to each other that a repeatability versus positioning error is possible. We find that the repeatability decreases by a factor of approximately 2 under an overburden lens. Furthermore, we find that the X‐ and Y‐components have approximately the same sensitivity to positioning errors as the Z‐component (for the same events) in this three‐dimensional vertical‐seismic‐profiling experiment. This indicates that in an area with complex overburden, positioning errors between monitor and base seismic surveys are significantly more critical than outside such an area. This study is based on a three‐dimensional three‐component vertical‐seismic‐profiling data set from a North Sea reservoir and care should be taken when extrapolating these observations into a general four‐dimensional framework.