Amyloidosis denotes the medical disorders associated with deposition of insoluble protein fibrillar aggregates and it is associated with various human diseases. Presence of aggregation prone regions plays an important role in determining the aggregation propensity of a protein, hence understanding the characteristics of these regions is of keen interest in academia and industry. In this work, we have identified 465 aggregation prone regions with 353 unique peptides in human proteome. Evaluation of the performance of available methods for identifying these 353 peptides showed a sensitivity in the range of 15% to 90%. Further, we identified the amino acid properties enthalpy, entropy, free energy and hydrophobicity are important for promoting aggregation. Utilizing these properties, we have developed a model for distinguishing between amyloid forming and non-amyloid peptides, which showed an accuracy of 71% with a balance between sensitivity and specificity. We suggest that the results obtained in this work could be effectively used to improve the prediction performance of existing methods. © Springer International Publishing AG 2017.