A series of laboratory column experiments were conducted in a glass bead medium with 36Cl and 85Sr as nonreactive and reactive tracers, respectively. The primary purpose was to compare breakthrough curve (BTC) properties under various conditions with those of the model which accounts for pore scale variation in the retardation factor (R) caused by pore scale heterogeneity. In particular, the purpose was to determine whether pore scale heterogeneity contributes to the nonideality of reactive solute BTCs that cannot be described by the conventional local equilibrium assumption (LEA)-based advection-dispersion equation (ADE) for a homogeneous medium. To eliminate other sources of nonideality, it was verified that the LEA was valid for the solute and porous medium used in the study. The following BTC properties were observed: (1) nonreactive BTCs can be described by the ADE (ideal), whereas reactive BTCs are nonideal, (2) materials with larger pore size variation gave greater nonideality in the reactive BTC, (3) although nonreactive BTCs showed equally symmetric ideal shapes regardless of distance, reactive BTCs showed slight travel distance dependence in the degree of nonideality, and (4) the position and shape of dimensionless BTCs are independent of mean fluid velocity. Calculated stochastic dispersivities which accounted for measured pore size variations were significantly larger than those for a nonreactive solute but were smaller than the effective dispersivities actually observed for the reactive BTCs. From these findings, it is concluded that pore scale variation in R caused by pore scale heterogeneity in the medium is at least partially responsible for the nonideality observed for reactive solute transport in homogeneous media. Pore scale variation in R gives larger asymptotic dispersivity and thus longer preasymptotic time and distance for reactive solute transport. ¿ American Geophysical Union 1995