A general approach is proposed that allows for quantifying the relative toxic contribution of ions released from metallic nanoparticles and of the particles themselves, as exemplified for the case of differently shaped zinc oxide (ZnO) nanoparticles (NPs) exposed to zebrafish embryos. First of all, the toxicity of suspensions of ZnO nanoparticles (NP(total))--nanospheres, nanosticks, cuboidal submicron particles (SMPs), and Zn(NO3)2--to the embryos was assessed. The observed toxicity of ZnO NP(total) is assumed to result from the combined effect of the particles present in the suspensions (NP(particle)) and of the dissolved Zn(2+) ions released from the particles (NP(ion)). Different addition models were used to explicitly account for the toxicity of NP(particle). The median lethal concentrations (LC50) of NP(particle) of nanospheres, nanosticks, and SMPs were found to range between 7.1 mg Zn/L and 11.9 mg Zn/L (i.e., to differ by a factor of 1.7). Behavioral performance showed no significant differences among all types of the NP(particle). The median effective concentrations (EC50) of the particles were found to range between 1.0 mg Zn/L and 2.2 mg Zn/L. At the LC50 of each particle suspension, the main contribution to lethality to zebrafish embryos was from the NP(particle) (52%-72%). For hatching inhibition, the NP(particle) was responsible for 38% to 83% of the adverse effects observed. The ZnO nanosticks were more toxic than any of the other NPs with regard to the endpoints mortality and hatching inhibition. The main contribution to toxicity to zebrafish embryos was from the NP(particle) at the LC50 and EC50 of each particle suspension.