Regeneration of spent caustic from liquefied petroleum gas sweetening in refineries is significant to save resources and reduce solid waste emissions. Conventional regeneration technologies have been widely used, but these technologies usually have a reactor and a separator in series which occupies a large space. In this work, simultaneous reaction and separation processes were proposed to be conducted in a rotating packed bed (RPB), aiming for the process intensification of spent caustic regeneration. Experimental results of laboratory tests show that the regeneration processes of mercaptide oxidation and disulfide separation were efficiently enhanced in the RPB. The side-line test, treating the actual spent caustic from a refinery, could achieve good performance for a long-time running. An artificial neural network model was applied for parameters analysis and prediction of the regeneration performance. This work demonstrated the feasibility of spent caustic regeneration in only one RPB unit, which displays bright prospects for industrial application.