This research study investigates pre-COVID-19 and COVID-19 student prediction retention models based on student background characteristics at a northeastern US college. The study employs logistic regression in developing models to compare pre-pandemic and pandemic prediction factors and prediction accuracy. Study results indicate that the pre-COVID-19 model factors of first term GPA, work study work amount and being a male student are negatively associated with withdrawal from school while unmet financial need is positively associated with withdrawal from school. Study results suggest pandemic model predictive factors of first term GPA and year 1 loans are negatively associated with withdrawal from school while unmet financial need is positively associated with withdrawal from school. The pandemic model is better in predicting student withdrawals than the pre-pandemic model. Study provides further knowledge for researchers, and college and university administrators in understanding student retention factors during normal and disruptive periods such as the pandemic.ULL