Rainbow trout, Oncorhynchus mykiss (Walbaum), were exposed continuously to infectious pancreatic necrosis virus (IPNV) at 0, 10(1), 10(3) or 10(5) plaque forming units (pfu) L(-1) of water to estimate the effects of chronic IPNV exposure on early life stages. Fish density averaged 35 fish L(-1) (low density) or 140 fish L(-1) (high density), and the tank flow rate was 250 mL(-1) min. Virus exposure began at 6 days before hatch and continued until fish were 44 days old. Cumulative per cent mortality, analysis of survival and hazard functions, and discrete-time event analysis were used to explore the patterns of survival and mortality. In eggs and fish exposed to IPNV, mortality significantly greater than in the 0 pfu L(-1) exposure did not occur until IPNV concentration was 10(5) pfu L(-1) at low fish density and 10(3) pfu IPNV L(-1) at high fish density. These results suggest that in the natural aquatic environment, where rainbow trout densities are likely to be considerably lower than in this study, mortality resulting from infection with IPNV will very likely not occur when ambient concentrations of virus are < or =10(3) pfu IPNV L(-1). In aquaculture rearing units, trout density is likely to be as high or higher than the densities used in this study. Therefore, continuous inputs of virus at concentrations greater than 10(1) pfu L(-1) may result in IPN epidemics in aquaculture facilities.