Sediment cores recovered during IODP Expedition 323 in the Bering Sea, northern Pacific, contained numerous ice-rafted debris (IRD) clasts up to 85 mm in length. The physical properties (including roundness and sphericity) of 136 clasts from the working half of the cores, a subsample of the total clast number, were analysed and their composition determined using standard petrographic techniques. After removal of pumice and possible fall-in derived material from the clast population, a total of 86 clasts from the original collection were considered to be IRD. While roundness and sphericity vary greatly in the clast population, the IRD are predominately discoid in shape with oblate/prolate indices typically between −5 and 5. There are four time periods over the approximately 4.5 Ma sample interval, 0.36-0.67 Ma, 0.82-1.06 Ma 1.54-1.77 Ma and >3.28 Ma, where there are no IRD in the sample set for sites of the Bering slope, suggesting that these times may have been ice-free. Most clasts show some rounding and are likely to have spent time on beaches with wave action. Wave action on beaches suggests periods of no ice or only seasonal sea-ice. The low roundness values of other clasts, however, suggest they underwent little working and, therefore, the presence of glaciers or more permanent sea-ice at times in those locations. The abundance of rounded and unfaceted clasts as IRD suggests a lack of large ice sheets in the area during cool periods. Clast composition of the IRD is divided into four broad groups, basalt and andesite, granite and metamorphic, sedimentary, and felsic volcanic. The granite and metamorphic and more mature sedimentary lithologies are most likely derived from the Alaskan continental margin, while the extrusive igneous clasts could be derived from a variety of volcanic sources surrounding the Bering Sea, both emergent now or emergent at times of lower sea level. There is only a poor correlation with IRD abundance and marine isotope stages (MIS) for the time period <1 Ma. Abundant IRD occurs in MIS 3 and can be correlated with MIS back to 400 Kyr but not to older ages. This suggests that the abundance of IRD >2 mm transported by sea-ice may not be a good indicator of past climate conditions.