An apparatus for assembling a head unit has a carriage, a recognition apparatus for image-wise recognizing a position of each of liquid droplet ejection heads, a moving apparatus for moving the carriage, a correction apparatus for slightly moving head holding members, a fixing apparatus for fixing each of the head holding members, and a controller. The controller drives the moving apparatus based on a result of recognition by the recognition apparatus, and then positions each of the liquid droplet ejection heads to the carriage, and then each of the liquid droplet ejection heads is fixed to the head holding members.