The elucidation of the molecular mechanisms by which soluble proteins convert into their amyloid forms is a fundamental prerequisite for understanding and controlling disorders that are linked to protein aggregation, such as Alzheimer's and Parkinson's diseases. However, because of the complexity associated with aggregation reaction networks, the analysis of kinetic data of protein aggregation to obtain the underlying mechanisms represents a complex task. Here we describe a framework, using quantitative kinetic assays and global fitting, to determine and to verify a molecular mechanism for aggregation reactions that is compatible with experimental kinetic data. We implement this approach in a web-based software, AmyloFit. Our procedure starts from the results of kinetic experiments that measure the concentration of aggregate mass as a function of time. We illustrate the approach with results from the aggregation of the β-amyloid (Aβ) peptides measured using thioflavin T, but the method is suitable for data from any similar kinetic experiment measuring the accumulation of aggregate mass as a function of time; the input data are in the form of a tab-separated text file. We also outline general experimental strategies and practical considerations for obtaining kinetic data of sufficient quality to draw detailed mechanistic conclusions, and the procedure starts with instructions for extensive data quality control. For the core part of the analysis, we provide an online platform (http://www.amylofit.ch.cam.ac.uk) that enables robust global analysis of kinetic data without the need for extensive programming or detailed mathematical knowledge. The software automates repetitive tasks and guides users through the key steps of kinetic analysis: determination of constraints to be placed on the aggregation mechanism based on the concentration dependence of the aggregation reaction, choosing from several fundamental models describing assembly into linear aggregates and fitting the chosen models using an advanced minimization algorithm to yield the reaction orders and rate constants. Finally, we outline how to use this approach to investigate which targets potential inhibitors of amyloid formation bind to and where in the reaction mechanism they act. The protocol, from processing data to determining mechanisms, can be completed in <1 d.