We studied the synchronism between the seasonal occurrence of fish larvae and their prey in ice-covered southeastern Hudson Bay, Canada, in spnng 1988, 1989 and 1990. Arctic cod Boreogadus saida and sand lance Ammodytes sp. larvae hatched several weeks before ice break-up and fed primarily on copepod nauplii. The timing of 50% yolk resorption was the same every year (1 1 to 18 May for Arctic cod and 5 to 11 June for sand lance) but the availability of copepod naupli~ varied substantially between years, both in magnitude (7-fold) and timing (4 to 6 wk), lnterannual differences in the under-ice abundance of nauplii were linked to variations in the abundance of female cyclopoid copepods, and appeared unrelated to the timing of the ice-algal or phytoplankton blooms. Interannual differences (2-to 4-fold) in the feeding success of fish larvae (percent feeding incidence at length and mean feeding ratio at length) were related to the availability of copepod nauplii Consistent with the match/mismatch hypothesis, the fixity of the spawning season in relation to a variable cycle of prey abundance accounted for the observed variations in feeding success and apparent growth (length at date) of fish larvae. Yet, in this particular ecosystem, a match or mismatch between Arctic cod or sand lance larvae and their prey may depend more on the dynamics of cyclopoid copepods during the previous winter than on the timing of the spring algal blooms.