Purpose: While a pathologic complete response (pCR) is regarded as a surrogate endpoint for positive outcomes in breast cancer (BC) patients receiving neoadjuvant chemotherapy (NAC), forecasting the prognosis of non-pCR patients is still an open issue. This study aimed to create and evaluate nomogram models for estimating the likelihood of disease-free survival (DFS) for non-pCR patients. Methods: A retrospective analysis of 607 non-pCR BC patients was conducted (2012–2018). After converting continuous variables to categorical variables, variables entering the model were progressively identified by univariate and multivariate Cox regression analyses, and then pre-NAC and post-NAC nomogram models were developed. Regarding their discrimination, accuracy, and clinical value, the performance of the models was evaluated by internal and external validation. Two risk assessments were performed for each patient based on two models; patients were separated into different risk groups based on the calculated cut-off values for each model, including low-risk (assessed by the pre-NAC model) to low-risk (assessed by the post-NAC model), high-risk to low-risk, low-risk to high-risk, and high-risk to high-risk groups. The DFS of different groups was assessed using the Kaplan–Meier method. Results: Both pre-NAC and post-NAC nomogram models were built with clinical nodal (cN) status and estrogen receptor (ER), Ki67, and p53 status (all p < 0.05), showing good discrimination and calibration in both internal and external validation. We also assessed the performance of the two models in four subtypes, with the triple-negative subtype showing the best prediction. Patients in the high-risk to high-risk subgroup have significantly poorer survival rates (p < 0.0001). Conclusion: Two robust and effective nomograms were developed to personalize the prediction of DFS in non-pCR BC patients treated with NAC.