Aimed at the problem that communication bandwidth is limited in wireless sensor networks, a distributed Kalman consensus filtering algorithm is designed based on eventtriggered strategy. The transmission mechanism of sendondelta is adopted. Each sensor sends its own observations to the corresponding estimators only when the square of difference between the current observation and the latest sent observation exceeds a tolerable threshold. In addition, each estimator can receive estimates from its neighbor nodes through timetriggered rules. In order to avoid the calculation of crosscovariance matrices between estimators, an eventtriggered distributed Kalman filter is proposed by locally minimizing an upper bound of the variance. The exponential boundedness of the algorithm in the sense of mean square is proved by the Lyapunov method. The numerical simulation shows that the less the eventtriggered threshold value, the greater the communication rate, and the higher the estimation accuracy of the proposed filter. Otherwise, the lower the communication rate, the lower the estimation accuracy.