An experiment was conducted to examine the efficacy of vapor extraction to remove volatile constituents from a floating Light Non-aqueous Phase Liquid (LNAPL) layer established in a tank without introduction of residual phase hydrocarbons in the vadose zone. Vapor extraction well design considerations were examined by varying the height of the intake manifolds lowest port. The experiment was conducted in two phases, the first of which was presented previously in 1996. During this phase of the experiment (Phase II), the vapor extraction was re-initiated in the tank to examine several questions still remaining after the first phase was completed. These included examination of long-term removal trends from the LNAPL, examination of system efficiency at different heights above the LNAPL, examination of constituent distributions in the tank at the end of the experiment, and evaluation of possible mechanisms controlling hydrocarbon removal from the LNAPL zone. This phase of investigation consisted of three main parts. Initial conditions of vapor and LNAPL in the tank were evaluated to establish continuity between Phase I and Phase II. After the vapor extraction was re-initiated and equilibrium conditions had been re-established in the tank, ports of the extraction manifold were sequentially closed to evaluate how system design changed the mass flux rates. Long-term monitoring was conducted to examine if any changes would be observed long-term removal rates. Finally, hydrocarbon concentrations in the tank were evaluated as part of an overall mass balance and to examine constituent distributions in the LNAPL zone at the end of the experiment. Several conclusions have been found for this phase of investigation. First, several processes appear to control removal of gasoline range organic constituents from an LNAPL layer. Mass removal of the gasoline constituents of LNAPL initially may be controlled by either direct vaporization or partitioning from dissolved phase, which is saturated with respect to each constituent as air flow is directed through the enhanced LNAPL capillary fringe formed near the extraction well and the zones of highest hydrocarbon saturation at the air/LNAPL interface. Removal rates decrease some point after the start of the vapor extraction system. Results from mass balance modeling suggest this decrease in removal rates could result from diffusion limiting the mass transfer as the outer zones of hydrocarbons begin to saturate the air flow entering the extraction well, as would occur after the enhanced LNAPL capillary fringe has been depleted of hydrocarbons. The decreases in mass removal rates observed later in the experiment may be due to diffusion from smaller zones of residual gasoline or to partitioning from the dissolved phase, which no longer exists at concentrations near the effective solubility. Second, design considerations are extremely important to system efficacy. Remediation of LNAPL is possible, but system design requires that well screen be placed close to the top of the LNAPL/air interface. System design would depend on the degree of upwelling and formation of an enhanced LNAPL capillary fringe. Third, removal of constituents during vapor extraction occurs first from areas away from the extraction well and progressively moves towards the extraction well. Fourth, heterogeneities within the screened soil column do not significantly affect mass flux removal rates. Finally, removal of constituents found in gasoline from an LNAPL layer is possible using soil vapor extraction, but age time that the LNAPL has been above groundwater is important to compound removal efficacy, especially extremely soluble constituents such as MTBE.