Abstract: Objective To explore the learning curves of simulation-based lumbar puncture training and evaluate its long-term outcome,which can provide a basis for arrangements and effect assessments of simulated training for medical undergraduates. Methods Twenty-five medical undergraduates who had not received any relevant lumbar puncture training and had no clinical experience of lumbar puncture were enrolled in this study.Each participant performed seven supervised trials on the simulator in two weeks.Participants’ performances were assessed by operation mistake(OM),operation time(OT) and operation confidence(OC).Learning curves for each variable were generated.Long-term outcome of the simulation-based lumbar puncture training was measured six months later after the initial training on the simulator. Results Obvious improvements were noted in OM,OT,and OC among the first four trials(P＜0.05).The learning curves of OM and OT descended in the first five trials.However,the learning curve of OC ascended in the first five trials.Plateaus of OM,OT,and OC in the learning curves occurred in the fifth trial.There was no significant difference between the result of the retest performed six months later and that of the seventh trial(P＞0.05). Conclusion This study demonstrates that simulation-based lumbar puncture training can significantly improve the individual’s skill.The learning curves of OM and OT descend with the numbers of operation,while the learning curve of OC ascends with the numbers of operation,but they all reach plateaus in the fifth operation.Good effect can be achieved by five trials.