An artificial neural network model was developed to correlate the relationship between the alloying elements (Al, V, Fe, O, and N) and heat treatment temperature (inputs) with the volume fractions of α and β phases (outputs) in some α, near-α, and α + β titanium alloys. The individual and combined influences of the composition and temperature on α and β phases were simulated through performing sensitivity analysis. A new method has been proposed to estimate the relative importance of the inputs on the outputs for single phase α-Ti, near-α Ti, and α + β Ti alloys. The average error of the model predictions for 35 unseen test data sets is 1.546%. The estimated behavior of volume fractions of α and β phases as a function of composition and temperature are in good agreement with the experimental knowledge. Justification of the results from the metallurgical interpretation has been included. © 2015 Elsevier B.V. All rights reserved.