The potential toxicity of nanoparticles is addressed by utilizing a putative attractive model in developmental biology and genetics: the zebrafish (Danio rerio). Transparent zebrafish embryos, possessing a high degree of homology to the human genome, offer an economically feasible, medium-througput screening platform for noninvasive real-time assessments of toxicity. Using colloidal silver (cAg) and gold nanoparticles (cAu) in a panoply of sizes (3, 10, 50, and 100 nm) and a semiquantitative scoring system, it is found that cAg produces almost 100% mortality at 120 h post-fertilization, while cAu produces less than 3% mortality at the same time point. Furthermore, while cAu induces minimal sublethal toxic effects, cAg treatments generate a variety of embryonic morphological malformations. Both cAg and cAu are taken up by the embryos and control experiments, suggesting that cAg toxicity is caused by the nanoparticles themselves or Ag+ that is formed during in vivo nanoparticle destabilization. Although cAg toxicity is slightly size dependent at certain concentrations and time points, the most striking result is that parallel sizes of cAg and cAu induce significantly different toxic profiles, with the former being toxic and the latter being inert in all exposed sizes. Therefore, it is proposed that nanoparticle chemistry is as, if not more, important than specific nanosizes at inducing toxicity in vivo. Ultimately such assessments using the zebrafish embryo model should lead to the identification of nanomaterial characteristics that afford minimal or no toxicity and guide more rational designs of materials on the nanoscale.