In this study, lung tumour motion modelling and prediction was done using a fuzzy logic approach. The surrogate signal for breathing motion was obtained from 10 different instances of a lung patient by using signals from Realtime Respiratory Motion Management (RPM) system. A regularity criterion (RC criterion) was used to select appropriate inputs for the model for each trace. A first order Sugeno type model was devised by using a subtractive clustering approach. On an average, the prediction error was seen to be 0.21 mm for training and 0.23 mm for testing. The two main advantages of using a fuzzy logic approach are reduction in the input data to enable faster model building and linguistic interpretability for easier clinical use.