In this paper, significant variables of domestic urban water demand required for the purpose of estimation of urban water supply in five planned colonies of the City of Ajmer, Rajasthan, India, are identified. The data for these 16 variables are entered in the multiple linear regression (MLR) (stepwise) models in SPSS software, and domestic water demand models are developed. Based on these models, the six most significant variables, namely temperature (T), rainfall (RF), family size (FS), family income (FI), number of bathrooms (NB), and age of house (AH), are identified. The data of 16 variables are further utilized in principal component analysis (PCA), and five factors/variables are extracted, comprising combinations of these 16 variables. A regression coefficient of 0.76 is obtained in the PCA model. These six significant variables are further fed into a multilayer perceptron neural network (NN) model for water demand forecasting. The linear regression coefficient of NN is 0.90, very close to the MLR (stepwise) coefficient of 0.89, and verifying the dependence of water demand on these six variables. The outcome of the study suggests that the six variables are significant for estimation of water demand for Ajmer.