Detailed Cluster observations of flank magnetopause reconnection are presented for two events on the Northern and the Southern Hemispheric dawnside flanks when the interplanetary magnetic field (IMF) clock angle $theta$ = arctan(By/Bz) is within ~45¿ of the equatorial plane. The event selection is based on the relative proximity between the Cluster spacecraft 1 position and the predicted magnetospheric sash where antiparallel merging is expected to develop. MHD simulations performed for the two events indicate that the Cluster spacecraft were passing through the MHD sash region in the Northern Hemisphere on 30 June 2001, while crossing the magnetopause equatorward of the Southern Hemispheric sash on 29 May 2001. Accelerated and decelerated plasma flows relative to the magnetosheath velocity were detected by Cluster on both occasions. The Wal¿n test confirms that the observed ΔV is directly correlated with the predicted magnetic field rotation ΔB/$sqrt{rho mu_{0}}$ with the expected direction of the normal magnetic field and so we interpret them as speed changes due to magnetic reconnection. The observed directions of ΔV compare very well with the location of the simulated MHD sash relative to Cluster. The magnetic field shear in the locally tangential plane of the magnetopause ranges between 171¿ and 177¿ for the 30 June event in good agreement with antiparallel merging at the MHD sash. The corresponding local field shear for the 29 May event is only 144¿, either suggesting a component merging site in the direction of the sash or indicating that Cluster is farther away from the location where the neutral line was initially formed as compared with the 30 June event. A comparison between the projected regions of antiparallel and component merging onto the magnetopause and the quasi-steady direction of plasma acceleration detected by Cluster on 29 May and 30 June support the view that the IMF controls the expected global location of magnetic reconnection at limited regions of the magnetopause. This is in contrast to randomly distributed merging sites over the dayside magnetopause that in principle could generate accelerated flows away from the observed directions of ΔV.