A hybrid ensemble transform filter is introduced which combines a Kalman-filter update provided by the analysis scheme of the local ensemble transform Kalman filter (LETKF) and a nonlinear analysis following the nonlinear ensemble transform filter (NETF). The NETF computes an analysis ensemble from particle weights and has been shown that it can yield smaller errors than the LETKF for sufficiently large ensembles. The new hybrid filter is motivated from combining the stability of the LETKF with the nonlinear properties of the NETF to obtain improved assimilation results for smaller ensembles. The hybrid filter is domain-localized as the LETKF and can hence be applied to high-dimensional nonlinear models. Its performance depends on the choice of the hybridization weight which shifts the analysis solution between the LETKF and NETF analyses.