Reinforced concrete columns (RCCs) are crucial structural elements in buildings as they support all loads acting on the structure and transfer them to the foundation. During a fire, RCCs are not only subjected to eccentric loads but also exposed to high temperatures. Therefore, accurately estimating the fire resistance of RCCs under eccentric loads is essential to ensure building stability and minimize risks to people and property. To predict the fire resistance of RCCs under eccentric loads, two popular machine learning approaches, artificial neural network approach (ANNA), and random forest approach (RFA), were developed using a dataset of 350 fire test results. The dataset consisted of 12 input variables and 1 output variable. The prediction results demonstrated acceptable correlations between predicted and actual data in the training set, with correlation coefficients exceeding 0.94 and root mean squared errors below 33 min, which is less than 15% of the mean actual fire resistance values. However, both approaches showed a reduction in accuracy in the testing set. RFA slightly outperformed ANNA in predicting the fire resistance of RCCs in both the train and test sets. Furthermore, cross-sectional height emerged as the most crucial parameter impacting the fire resistance of RCCs under eccentric loads according to both approaches.