3 He concentrations were measured in zircon, apatite and pyroxene separates from rhyolite surfaces along the Snake River near Twin Falls, Idaho. These fluted and potholed surfaces were presumably scoured by the Bonneville outburst flood ∼17.5ka. Several of the samples contain inherited 3 He from prior exposure, reflecting the complex incision history of the Snake River canyon and suggesting earlier flood events. Each individual mineral from the remaining surfaces yielded 3 He concentrations that are within error of each other. 3 He produced by neutron capture on 6 Li was established from shielded samples and a simple neutron production model. The remaining 3 He is spallogenic and was used to calibrate the production rate of cosmogenic 3 He in zircon, apatite and pyroxene by assuming minimal erosion since the Bonneville flood. In all three phases the resulting production rates are ∼7–12% lower than previous 3 He production rate estimates obtained by reference to various production rates for 10 Be in quartz. This disagreement is partially reconciled here by recalculating previous results with a 10 Be production rate of 4.51 at g −1 a −1 in quartz. Adopting this revised rate brings three independent 3 He production rate estimates for zircon and apatite into agreement within 5%, with grand means of 103±3 at g −1 a −1 in zircon and 133±6 at g −1 a −1 in apatite. The major source of uncertainty in cosmogenic 3 He dating of these phases is now correction for 3 He produced by neutron capture on 6 Li. Calculations are presented for assessing the amount of uncertainty introduced by this correction as a function of Li concentration, cosmic-ray exposure age, and He closure age.