The sharp increase in production of engineered nanomaterials (ENMs) combined with their high potential for aquatic toxicity, mean that understanding the transport of these materials throughout the environment is of utmost importance. In the presence of environmentally relevant surfaces, the relationship between particle attachment and relevant variables was quantitatively investigated and reported. Increasing temperature greatly altered the attachment onto two different humic substances by altering the hydration, and therefore confirmation, of the natural organic matter (NOM) matrix. By increasing the hydration of the NOM layer, the matrix swelled, allowing for more surface area for particle attachment and an increase in possible sorption sites. Similarly, high ionic strengths caused the NOM layer to condense, reducing surface area and sorption sites for particle attachment and effectively lowering particle attachment efficiency. The shape of the particle itself also played a role in attachment. A humic acid layer showed preference to smaller, more spherical particles due to the size of the voids within the layer, raising attachment efficiency for the smaller, spherical particles only, while a smoother, more condensed layer did not. As ionic strength increased, however, the layer condensed and the preference vanished. Finally, a predictive model for attachment efficiency was developed using a machine learning approach and trained on a database containing all the data gathered in this work combined with all currently available, relevant attachment efficiency literature. The model employed 13 training features, each of which was a physicochemical characteristic of the particle, surface, or solution system, to predict attachment efficiency with relatively high performance. The most important features for predicting attachment efficiency were also identified. The results presented in this work improve the understanding of particle attachment efficiency by identifying important variables, explaining why these variables have an effect on attachment efficiency, and also providing an empirical predictive model for attachment efficiency. By applying this approach to other areas of particle transport, we can close the gap between experimental and modeling efforts, advancing transport knowledge as quickly and efficiently as possible. Only by closing this gap can we expect to understand particle transport in a system as complex as the natural environment.