Studies of cycad pollination globally have revealed obligate pollination mutualisms with insects, typically host-specific beetles. Some cycads in the genus Cycas have floating seeds allowing oceanic dispersal, and have colonized remote islands such as Fiji, Guam, and Madagascar. This seems to contradict the expected requirement that specific beetle pollinators be present in order for cycads to establish in remote localities. We therefore postulated that these ‘island’ cycads are wind pollinated. We tested for adaptations associated with wind pollination in a laboratory study of pollen aerodynamics. The pollen from a range of (non-cycad) wind-pollinated plants was not relatively ‘buoyant’: the settling velocities of wind-pollinated species exceeded those of insect-pollinated cycads and some animal-pollinated angiosperms. We attributed this result to settling velocity being, in part, correlated with pollen size, as cycad pollen was consistently smaller. A better predictor for pollination vector was the tendency of pollen to aggregate during a vertical fall. Pollen of wind-pollinated plants separated into single grains, whereas cycad pollen and pollen of animal-pollinated angiosperms adhered in large aggregations, the settling velocity of which increased rapidly with aggregate size. The pollen of cycads known to be insect pollinated was similar in morphology, settling speed, and clumping behaviour to pollen of ‘island’ cycads of unknown pollination biology. Based on these aerodynamics, we predict that cycad species on oceanic islands have insect pollinators. More broadly, we propose that pollen aggregation during a vertical fall provides a sound test for wind versus animal pollination when the pollination vector is unknown. © 2011 The Linnean Society of London, Biological Journal of the Linnean Society, 2011, 104, 75–92.