Non-piloted radiative ignition and transition to flame spread over thin cellulose fuel samples was studied aboard the USMP-3 STS-75 Space Shuttle mission, and in three test series in the 10 second Japan Microgravity Center (JAMIC). A focused beam from a tungsten/halogen lamp was used to ignite the center of the fuel sample while an external air flow was varied from 0 to 10 cm/s. Non-piloted radiative ignition of the paper was found to occur more easily in microgravity than in normal gravity. ignition of the sample was achieved under all conditions studied (shuttle cabin air, 21%-50% O-2 in JAMIC), with transition to flame spread occurring for all but the lowest oxygen and Bow conditions. Although radiative ignition in a quiescent atmosphere was achieved, the flame quickly extinguished in air. The ignition delay time was proportional to the gas-phase mixing time, which is estimated by using the inverse flow rate. The ignition delay was a much stranger function of Bow at lower oxygen concentrations. After ignition, the flame initially spread only upstream, in a fan-shaped pattern. The fan angle increased with increasing external flow and oxygen concentration from zero angle (tunneling flame spread) at the limiting 0.5 cm/s external air Bow, to 90 degrees (semicircular flame spread) for external Bows at and above 5 cm/s, and higher oxygen concentrations. The fan angle was shown to be directly related to the limiting air flow velocity. A surface energy balance reveals chat the heat feedback rate from the upstream flame to the surface decreases with decreasing oxygen mass transport via either imposed flow velocity or ambient oxygen concentration. Quenching extinction occurs when the heat feedback rate from the flame is no longer sufficient to offset the ongoing surface radiative heat losses. Despite the convective heating from the upstream flame. the downstream flame was inhibited due to the 'oxygen shadow' of the upstream flame for the air Bow conditions studied. Downstream flame spread rates in air, measured alter upstream flame spread was complete and extinguished, were slower than upstream flame spread rates at the same Bow. The quench regime for the transition to flame spread was skewed toward the downstream, because of the augmenting role of diffusion for opposed Bow flame spread, versus the canceling effect of diffusion at very low cocurrent flows.