An offshore platform is naturally vulnerable to accidents, such as the leakage of dangerous chemicals, fire and explosion. Oil and gas are explosive and all the equipment and pipes are squeezed into a limited area on a platform. Escape, Evacuation, and Rescue (EER) plans play a vital role as the last barrier to ensure the safety of personnel in the event of a major accident. As a result, the main contributors leading to evacuation failure need to be analyzed to prioritize technology development and select a robust EER strategy. This research aims to undertake the quantitative reliability analysis of various EER strategies on offshore platforms. First, a reliability prediction model of emergency evacuation is established for offshore platforms based on the K2 structure learning algorithm and a Bayesian network parameter learning method. The conditional probability table of each node is determined by combining the Bayesian estimation method and a junction tree reasoning engine. The reliability of emergency evacuation on a platform is predicted using a dynamic Bayesian network model. The transition probability is determined through a Markov method. The main factors leading to evacuation failure are investigated using the diagnostic reasoning method of Bayesian Network. The findings can provide insights for the development of cost effective EER strategies for an offshore platform.