A geomagnetic jerk is defined as a sudden change in the trend of the time derivative of geomagnetic secular variation. A statistical time series model is applied to monthly means of geomagnetic eastward component obtained at 124 geomagnetic observatories to detect geomagnetic jerks objectively. The trend component in the model is expressed by a second order spline function with variable knots. The optimum parameter values of the model including positions of knots are estimated by the maximum likelihood method, and the optimum number of parameters is determined based on the Akaike Information Criterion. The geomagnetic jerks are detected objectively and automatically by regarding the determined positions of knots as the occurrence epochs. This analysis reveals that the geomagnetic jerk in 1991 is a local phenomenon while the 1969 and 1978 jerks are confirmed to be global phenomena.