A series of unsaturated column experiments was conducted to evaluate light nonaqueous phase liquid (LNAPL) removal during surfactant solution infiltration. Surfactant-LNAPL phase behavior tests were conducted to optimize the remedial solutions. Packed sand and site sediment columns were first processed to establish representative LNAPL smear zones under unsaturated conditions. Infiltration of low-concentration surfactant was then applied in a stepwise flush mode, with 0.33 column pore volume (PV) of solution in each flush. The influence of infiltrated surfactant solution volume and pH on LNAPL removal was assessed. A LNAPL bank was observed at the front of the first surfactant infiltration in each column, indicating that a very low surfactant concentration is needed to reduce the LNAPL-water interfacial tension sufficiently to mobilize trapped LNAPL under unsaturated conditions. More LNAPL was recovered as additional steps of surfactant infiltration were applied. Up to 99% LNAPL was removed after six infiltration steps, with approximately 2.0 PV of total surfactant solution application, suggesting surfactant infiltration may be an effective method for vadose zone LNAPL remediation. The influence of pH tested in this study (3.99∼10.85) was insignificant because the buffering capacity of the sediment kept the pH in the column higher than the zero point charge (pHzp) of the sediment and therefore surfactant sorption was negligible.