One of the largest field studies of reactive-solute transport is the natural-gradient experiment conducted at Cape Cod from 1985 to 1988. Major findings regarding the transport behavior of the reactive solute (lithium) were that the rate of plume displacement decreased with time (temporal increase in effective retardation), the degree of longitudinal spreading was much greater than that observed for bromide for an equivalent travel distance, and the plume was asymmetric, with maximum concentrations located near the leading edges. The objective of our work was to quantitatively analyze the transport of lithium and to attempt to identify the factor or factors that contributed significantly to its observed nonideal transport. We used a mathematical model that accounted for several transport factors, including spatially variable hydraulic conductivity and spatially variable, nonlinear, rate-limited sorption, with all parameter values obtained independently. The transport behavior observed during the first 250 days, corresponding to a transport distance of 60 m, was predicted reasonably well by the simulation that incorporated spatially variable hydraulic conductivity; nonlinear, rate-limited, spatially variable sorption; and uniform water chemistry. However, the larger degree of deceleration observed during the latter stage of the experiment (the final 20 m) was not. The larger deceleration was successfully simulated by increasing 3-fold the mean sorption capacity of the latter portion of the transport domain. Such a change in sorption capacity is consistent with the potential impact on lithium sorption of measured changes in water chemistry (e.g., pH increase, reduction in resident Zn) that occur in the zone through which the lithium plume traversed. The results of the analyses suggest that nonlinear sorption and variable water chemistry may have been primary factors responsible for the nonuniform displacement of the lithium plume, with rate-limited sorption/desorption having minimal impact. In addition, the asymmetry of the plume appears to have been caused primarily by nonlinear sorption, whereas the enhanced longitudinal spreading appears to have been caused by the combined influences of spatially variable hydraulic conductivity and sorption, nonlinear sorption, and rate-limited sorption/desorption. A comparison of the results of this analysis to those we obtained from an analysis of the Borden natural-gradient study reveals several similarities regarding the transport of reactive contaminants at the field scale. ¿ 1999 American Geophysical Union