Recent studies suggest that continental flood basalts may be suitable for geologic carbon sequestration due to fluid-rock reactions that mineralize injected CO₂ on relatively short time-scales. Flood basalts also possess a permeability structure favorable for injection, with alternating high-permeability (flow margin) and low-permeability (flow interior) layers. However, little information exists on the behavior of CO₂ as it leaks through fractures characteristic of the flow interior, particularly at conditions near the critical point for CO₂. In this study, a two-dimensional 5 × 5 m model of a fracture network is built based on high-resolution LiDAR scans of a Columbia River Basalt flow interior taken near Starbuck, WA. Three-phase CO₂ flow is simulated using TOUGH3 (beta) with equation of state ECO2M for 10 years simulation time. Initial conditions comprise a hydrostatic pressure profile corresponding to 750-755 m below ground surface and a constant temperature of 32° C. Under these conditions, the critical point for CO₂ occurs 1.5 meters above the bottom of the domain. Matrix permeability is assumed to be constant, based on literature values for the Columbia River Basalt. Fracture permeability is assigned based on a lognormal distribution of random values with mean and standard deviation based on measured fracture aperture values and in situ permeability values from literature. In order to account for fracture permeability uncertainty, CO₂ leakage is simulated in 50 equally probable realizations of the same fracture network with spatially random permeability constrained by the lognormal permeability distribution. Results suggest that fracture permeability uncertainty has some effect on the distribution of CO₂ within the fractures, but network geometry is the primary control in determining flow paths. Fracture permeability uncertainty has a larger influence on fluid pressure, and can affect the location of the critical point within ~1.5 m. Uncertainty in fluid pressure was found to be highest along major flow paths below channel constrictions, indicating permeability at a few key points can have a large influence on fluid pressure distribution.