Over the next century, the Arctic is projected to become seasonally sea ice-free. Assessing feedback between clouds and sea ice as the Arctic loses sea ice cover is important because of clouds' radiative impacts on the Arctic surface. Here we investigate present-day and future Arctic cloud-sea ice relationships in a fully coupled global climate model forced by business-as-usual increases in greenhouse gases. Model evaluation using a lidar simulator and lidar satellite observations shows agreement between present-day modeled and observed cloud-sea ice relationships. Summer clouds are unaffected by sea ice variability, but more fall clouds occur over open water than over sea ice. Because the model reproduces observed cloud-sea ice relationships and their underlying physical mechanisms, the model is used to assess future Arctic cloud-sea ice feedback. With future sea ice loss, modeled summer cloud fraction, vertical structure, and optical depth barely change. Future sea ice loss does not influence summer clouds, but summer sea ice loss does drive fall cloud changes by increasing the amount of sunlight absorbed by the summertime ocean and the latent and sensible heat released into the atmosphere when the Sun sets in fall. The future fall boundary layer deepens and clouds become more opaque over newly open water. The future nonsummer longwave cloud radiative effect strengthens as nonsummer cloud cover increases. In summary, we find no evidence for a summer cloud-sea ice feedback but strong evidence for a positive cloud-sea ice feedback that emerges during nonsummer months as the Arctic warms and sea ice disappears. ©2018. American Geophysical Union. All Rights Reserved.