We analyzed ichthyoplankton data from six cruises off southwest Nova Scotia, spanning the spawning and early life history period (February-June), to determine the stage(s) at which the apparent failure of the 1983 haddock year class occurred. Egg production over the season did not differ significantly from that expected, but the abundance of larvae was 1–2 orders of magnitude lower than in recent years. Instantaneous egg mortality rates (M) calculated from catch curves were very high: 0.21–0.54 for the six cruises. The lowest egg mortality rates (M = 0.21–0.22) occurred during peak spawning (April), but no surviving larvae were obtained from this period, which indicates sharply increased mortality at or near the point of hatch. There appeared to be a weak relationship between wind displacement and diffusion rate of the egg mass, but drift rates were low, and wind speeds generally below average during the study period. Temperatures in the upper 50 m over the spawning ground, where the eggs predominantly occur, were in the normal range for haddock embryonic development throughout the study period. The dominant invertebrate predator encountered on the cruises was the ctenophore Pleurobrachia pileus, but calculations indicate it did not contribute significantly to egg mortality. Our study does not support the hypothesis that there is a single critical period in the early life history of haddock. The failure of the 1983 Browns Bank year class seems to have been determined prior to the period of first-feeding by larvae and to have been compounded over several developmental stages.