A temporal point process is a stochastic process that predicts which type of events is likely to happen and when the event will occur given a history of a sequence of events. There are various examples of occurrence dynamics in the daily life, and it is important to train the temporal dynamics and solve two different prediction problems, time and type predictions. Especially, deep neural network based models have outperformed the statistical models, such as Hawkes processes and Poisson processes. However, many existing approaches overfit to specific events, instead of learning and predicting various event types. Therefore, such approaches could not cope with the modified relationships between events and fail to predict the intensity functions of temporal point processes very well. In this paper, to solve these problems, we propose a variational neural temporal point process (VNTPP). We introduce the inference and the generative networks, and train a distribution of latent variable to deal with stochastic property on deep neural network. The intensity functions are computed using the distribution of latent variable so that we can predict event types and the arrival times of the events more accurately. We empirically demonstrate that our model can generalize the representations of various event types. Moreover, we show quantitatively and qualitatively that our model outperforms other deep neural network based models and statistical processes on synthetic and real-world datasets.