Agglomerated foods have gained attention in recent years due to their convenience in use. Turmeric powder has been subjected to agglomeration process at different moisture contents (10-28%) and steaming times (0-60 min). Experimental cumulative particle size distribution data of agglomerated samples can be predicted well (0.951 <= r <= 0.999, p <= 0.01) with Rosin-Rammler-Bennett model. The functional properties related to hydration characteristics like wetting time (10-35 s) and sinking time (15-115 s) of agglomerated samples decrease with an increase in moisture content and/or steaming time. Microstructural observation shows that the non-agglomerated sample possesses spheroids and ellipsoids of different sizes. The size of agglomerates ranges between 50 and 160 mu m; their shape varies from spheroid to elongated ellipsoids. Image analysis infers that the size related parameters increase with an increase in moisture content/steaming time. A four-layered artificial neural network having a structure of 2-10-8-4 has been developed to predict the agglomeration process of turmeric powder. (C) 2013 Elsevier Ltd. All rights reserved.