Breastfeeding intention is one of the strongest predictors of breastfeeding behavior and practice. The Theory of Planned Behavior (TPB), with its main construct, behavioral intention, is useful to predict actual behavior. While the literature has examined the implications of other theoretical notions such as self-identity, moral norm, descriptive norm, and socio-demographic variables, their roles remain unclear. Similarly, research on ethnic and low-income mothers is even insufficient. Therefore, given the original TPB constructs, our goal was to examine the role of extra theoretical constructs and specific demographic variables, and observe whether the original model would alter. A cross-sectional analytical study was carried out among 325 pregnant mothers in six purposively selected health care facilities of Kaski and Tanahu districts, using structured pro forma, from December 2018 to November 2019. We developed stepwise multivariate logistic regression from the entered and cleaned data, observed the effects on breastfeeding intention (BFI), and checked against multiple parameters. We interpreted the model with adjusted odds ratios and β coefficients, along with the variance explained. Out of 325 pregnant women, more than half (54.8%) were primiparous, and the mean age was 25.1±5 years. All three theoretical constructs of TPB regressed the BFI significantly, with 10.7%, the breastfeeding attitude dominating (β = 0.734, p = .003), and the other two constructs-perceived breastfeeding control (β = 0.659, p = .011) and breastfeeding subjective norm (β = 0.504, p = .045). Interestingly, breastfeeding self-identity added further variance of 5.2% (β = 0.955, p < .001), followed by income, with 3.2% (β = 0.856, p = .001), both of which also tallied large effect sizes. The model parameters showed consistency and robustness. We discuss the variables of the theoretical and extended model for BFI as well as the arguments for the explained variance. Self-identity is a strong and independent predictor of breastfeeding intention and, therefore, should be considered to better prepare for breastfeeding, especially among ethnic pregnant mothers. The direct and intermediate effects of self-concept and income need further study with more robust designs.