The presence of disinfection by-products (DBPs) releasing from ballast water management systems (BWMS) can cause a possible adverse effects on humans. The objectives of this study were to compute the Derived No Effect Levels (DNELs) for different exposure scenarios and to compare these levels with the exposure levels from the measured DBPs in treated ballast water. The risk assessment showed that when using animal toxicity data, all the DNELs values were approximately 103–1012 times higher than the exposure levels of occupational and general public exposure scenarios, indicating the level of risk was low (risk characterization ratios (RCRs) < 1). However, when using human data, the RCRs were higher than 1 for dichlorobromomethane and trichloromethane, indicating that the risk of adverse effects on human were significant. This implies that there are apparent discrepancies between risk characterization from animal and human data, which may affect the overall results. We therefore recommend that when appropriate, human data should be used in risk assessment as much as possible, although human data are very limited. Moreover, more appropriate assessment factors can be considered to be employed in estimating the DNELs for human when the animal data is selected as the dose descriptors.