Experimental toxicology studies for the purposes of setting occupational exposure limits for aerosols have drawbacks including excessive time and cost which could be overcome or limited by the development of computational approaches. A quantitative, analytical relationship between the characteristics of emerging nanomaterials and related toxicity is desired to better assist in the subsequent mitigation of toxicity by design. Quantitative structure activity relationships (QSAR's) and meta-analyses are popular methods used to develop predictive toxicity models. A meta-analysis for investigation of the dose-response and recovery relationship in a variety of engineered nanoparticles was performed using a clustering-based approach. The primary objective of the clustering is to categorize groups of similarly behaving nanoparticles leading to the identification of any physicochemical differences between the various clusters and evaluate their contributions to toxicity. The studies are grouped together based on their similarity of their dose-response and recovery relationship, the algorithm utilizes hierarchical clustering to classify the different nanoparticles. The algorithm uses the Akaike information criterion (AIC) as the performance metric to ensure there is no overfitting in the clusters. The results from the clustering analysis of 2 types of engineered nanoparticles namely Carbon nanotubes (CNTs) and Metal oxide nanoparticles (MONPs) for 5 response variables revealed that there are at least 4 or more toxicologically distinct groups present among the nanoparticles on the basis of similarity of dose-response. Analysis of the attributes of the clusters reveals that they also differ on the basis of their length, diameter and impurity content. The analysis was further extended to derive no-observed-adverse-effect-levels (NOAEL's) for the clusters. The NOAELs for the “Long and Thin” variety of CNTs were found to be the lowest, indicating that those CNTs showed the earliest signs of adverse effects.