Emotion recognition based on EEG have gained extensive popularity and been widely investigated. However, in previous works, a major assumption accepted by default is that the training and testing datasets share the same distribution. Unfortunately, this assumption is mostly invalid in real-world applications for the variation of EEG can cause the distribution discrepancy between datasets easily, which results in performance degeneration of traditional emotion recognition methods. To address this problem, we construct a novel joint adaptation networks (JAN) for emotion recognition with EEG variation. Extensive experimental evaluations through two EEG datasets demonstrate its validity, and further comparisons with the state of the arts methods also validate its superiority.