Early prediction of nurses’ turnover and turnover intention is essential to enhancing staff retention, ensuring job satisfaction, and maintaining the quality of patient care. This systematic review evaluated studies that used machine learning techniques to predict either actual nurse turnover or turnover intention, with the goal of identifying key predictive variables and assessing model performance. A comprehensive search was conducted across PubMed, CINAHL, Cochrane Library, PsycINFO, and Google Scholar, following PRISMA guidelines. Out of 596 records screened, eight studies met the inclusion criteria. These studies were appraised using the CASP Clinical Prediction Rule Checklist. The most frequently reported predictors were salary and age. While several models, such as Decision Tree and Random Forest, demonstrated high internal predictive accuracy, external validation was lacking across all studies, limiting generalizability. Future research should focus on validating models in diverse populations and healthcare settings and on improving standardization in outcome measures and reporting practices to enhance the applicability of predictive models in nursing workforce planning.