Abstract: Dispersivity, a property parameter defined to characterize the degree of dispersion and mixing, is crucial for the prediction and remediation of groundwater contamination. However, in field tracer tests, the real mixing effect has often been ignored. In this paper, an aquifer with horizontal layered structure was simulated by laboratory experiments. Three kinds of media were chosen to reconstructed experimental aquifer. Two monitoring methods, buried sensor and sensor in the well, were adopted to compare the difference of breakthrough curves, so as to investigate the influence of mixing effect in the well on the dispersion scale dependence. The results showed that mixing effect made breakthrough curve increase in a gradient manner with significant trailing phenomenon. The results also showed that mixing effect led to an overestimation of the dispersivity, when the advection-dispersion equation was used. The difference between the measured dispersivity and the actual dispersivity increases with the solute transport distance. Moreover, scale-dependency of dispersivity was observed in both sensor buried in the media and sensor in the well, and mixing effect amplified the scale effect of dispersivity significantly, which can provide references for evaluation and prediction of contaminant transport.