A short-term, pulse injection tracer experiment conducted in fractured quartzitic sandstone at Kukuan, Taiwan was analyzed. Tracer transport at the test site was dominated by advection but a specific attenuation mechanism leading to breakthrough curve (BTC) tailing also seemed to exist. Matrix diffusion was hypothesized as the transport mechanism that results in the tailing. This hypothesis was proved by comparing the field BTC with numerical simulation results obtained by the general-purpose flow/transport simulator, TOUGH2, based on a single-fracture conceptual model. Due to the lack of accuracy of estimating the interporosity flux by the conventional double porosity model (DPM), TOUGH2 was incorporated with the multiple interacting continua (MINC) scheme to simulate the transient characteristics of the interporosity flux. In MINC, rock matrix is discretized as a series of continua according to the perpendicular distance from the fracture that adjoins the matrix. The closer the rock matrix is to the fracture, the finer the rock matrix is discretized. This concept is fundamentally different from DPM in that rock matrix is no longer treated as a single continuum. Simulation results by TOUGH2-MINC have successfully reproduced the observed BTC tailing even under the dominating advection effect. Sensitivity studies showed that TOUGH2-MINC is sensitive to parameters including fracture aperture (2b), matrix porosity (n_m) and effective molecular diffusion coefficient in matrix (D_m). If 2b, n_m , D_m , are respectively 200 µm, 2%, 10^-11 m² s ^-1, and if hydrodynamic dispersion coefficient (D) is 1.69 × 10^-6 m² s^-1, TOUGH2-MINC result can well fit the field BTC. Furthermore, the importance of matrix diffusion was verified by fitting the field BTC with analytical solutions that either neglect matrix diffusion or consider the mass exchange between mobile and immobile zones within the fracture as the attenuation transport mechanism. It was found that the BTC tailing can only be modeled if matrix diffusion is taken into count. Therefore, matrix diffusion can explain BTC tailing for advection-dominated tracer transport in fractured sandstone.