Two peaks of eclosion dates, about 15 days apart, occur in the same batch of fertilized eggs in the marine midge, Pontomyia oceana. Two hypotheses, the variable adaptive peaks and the bet-hedging hypotheses, were proposed as the ultimate factor of the polymorphic phenomenon. They were tested by experiments controlling feeding amount and photoperiod, as well as selective breeding experiments. The offspring eclosing in the two peaks do not differ in fecundities, egg diameters, thorax and head lengths of males; this is not compatible with the variable adaptive peaks hypothesis. Both peaks exist under various feeding and photoperiods, although peak ratios differed in the former. The results in the first peak lineage did not support there is genetic component in peak ratio determination. The experiments in the second peak lineage had much lower success rates, although the results seemed to suggest a genetic component. The results in a more extreme selection experiment did not support that there is genetic component either. The present results are more compatible with the bet-hedging hypothesis. Wind velocity may be a factor hard to predict by the midges, and it may cause reproductive failure of them. Whereas high emergence synchronization, a prominent feature of the marine midge, may have advantages in many aspects, it also concentrates the risk of total reproductive failure. Spreading offspring to more than one suitable eclosion peak, the midge may have sacrificed short-term reproductive rate for long-term fitness.