Zones of dense, nonaqueous phase liquids (DNAPLs) are difficult to cha
racterize as to their volume, composition, and spatial distribution us
ing conventional ground-water extraction and soil-sampling methods. Su
ch incompletely characterized sites have negative consequences for tho
se responsible for their remedial design, e.g., the uncertainties in t
he optimal placement of ground-water extraction wells and in the durat
ion of remediation. However, the recent use of the partitioning interw
ell tracer test (PITT) to characterize DNAPL zones at sites in New Mex
ico [unsaturated alluvium] and in Ohio, Texas, and Utah [saturated all
uvium] demonstrates that the volume and spatial distribution of residu
al DNAPL can be determined with accuracy. The PITT involves injection
of a suite of tracers which reversibly partition to different degrees
between the DNAPL and the ground water or soil air resulting in the ch
romatographic separation of the tracer signals observed at the extract
ion well(s). The design of a PITT requires careful consideration of th
e hydrostratigraphic, hydraulic, and certain geochemical properties of
the alluvium being tested. A three-dimensional, numerical model of a
heterogeneous alluvial aquifer containing DNAPL has been developed for
use with the UTCHEM simulator to demonstrate partitioning tracer test
ing and to address questions that are frequently raised in its applica
tion. The simulations include (1) the estimation of DNAPL volume for t
he simple case where only residual DNAPL is present in heterogeneous a
lluvium, (2) sensitivity studies to demonstrate the effect of increasi
ngly low residual DNAPL saturation on the tracer signal, and (3) the e
ffect of free-phase DNAPL on the estimation of the volume of DNAPL pre
sent. Furthermore, the potential interference of sedimentary organic c
arbon as a DNAPL surrogate on the tracer signal is considered and show
n to be readily resolved by the careful choice of tracers. Finally, a
protocol for the use of PITTs in alluvial aquifers is presented.