<p>Temporal knowledge graph (TKG) event prediction is a crucial task that predicts elements of future events based on historical event data of TKG. Existing TKG event prediction methods restrict search space and time range to avoid huge computational consumption, resulting in a decrease in accuracy. In order to improve the accuracy and efficiency of TKG event prediction, a model CMPH (Combination Model of Paths and History) is proposed, which consists of a path memory module and a history memory module. The former finds the paths in advance by a TKG path search algorithm and learns to memorize the recurrent pattern for reasoning, preventing path search at inference stage. The latter adopts efficient encoder–decoder architecture to learn the features of historical events, which can avoid tackling a large number of structural dependencies and increase the inference efficiency. These two modules are combined by a gate component to jointly predict TKG events. Extensive experiments on various real-world datasets demonstrate that the proposed model obtains substantial performance and efficiency improvement for TKG event prediction task. Especially, it achieves up to 6.9% and 7.3% improvements in MRR and hit@1, respectively, and up to 21 times speedup at inference stage comparing to the state-of-the-art baseline.</p>