Childhood cancer survivors (CCSs) have a substantial risk of experiencing serious and potentially life-threatening late effects. Late effects are health complications of either the cancer or the treatment received. Significant improvements in childhood cancer treatment have led to an increase in the number of CCSs, which focused attention on the impact of late effects on CCSs’ health-related quality of life (HRQoL). Applying machine learning algorithms to predict the risk severity of late effects of CCSs could assist in long-term follow-up care planning and potentially improve health outcomes. This study analysed a dataset comprising 713 rhabdomyosarcoma survivors, followed by applying clustering- and classification algorithms to select the best approach for predicting the risk severity of late effects. Three clustering algorithms, including agglomerative hierarchical clustering (AHC), k-means, and density-based spatial clustering of applications with noise (DBSCAN), were independently applied to a subset of this dataset to cluster the survivors based on the severity of late effects. The results revealed that the k-means model outperformed the AHC and DBSCAN models. Following clustering, the performance of six classification algorithms - comprising a decision tree, random forest, support vector machine, logistic regression, gradient boosting, and multilayer perception, was evaluated to predict the risk severity of late effects of these survivors. The gradient boosting model was the top-performing classification model based on the evaluation of the performance metrics. This model ranked age at follow-up, age at diagnosis, neck radiotherapy, educational status, and head radiotherapy as the top five important features for predictions. This research is significant because, to the best of the authors’ knowledge, this is the first study to apply machine learning algorithms on a rhabdomyosarcoma survivor cohort to predict the risk severity of late effects.