The �iron fertilisation� hypothesis suggests controversially that atmospheric CO2 has been influenced by transport of iron-containing dust to the ocean surface1-3. Experiments in the Southern Ocean show that productivity, and subsequent CO2 drawdown, are enhanced by iron additions4. A carbon cycle model (forced by large values of Southern Ocean dust flux) indicates productivity changes during past glacial times could have reduced atmospheric CO2 by ~40ppm 5. However, Southern Ocean dust flux is very low at present and was increased, but still low, during past glaciations. Thus, as for the equatorial Pacific 1, 6, 7, distally-supplied, upwelled iron may be more significant than local dust-borne iron. Hence, Northern, not Southern, hemisphere dust may drive Southern Ocean productivity8,9. Here, we examine the flux and timing of N. Atlantic dust inputs in relation to the Vostok climate record. For the penultimate glaciation, two N. Atlantic dust peaks occurred. At 155 ka, the Atlantic dust flux was 2500x that at Vostok10, but declined well before the onset of the Vostok CO2 rise. The second dust peak, at 130ka, substantially post-dated the CO2 rise. Thus, low Southern Ocean dust fluxes, and this mismatch between N. hemisphere dust peaks and Southern Ocean climate change, appear not to support the suggested role of dust-mediated iron fertilisation in the Southern Ocean at Termination II.