PURPOSE: Light scatter results in degradation of visual function. An optical bench model was used to identify the origins of scatter in the setting of a Boston keratoprosthesis (KPro). The effect of various modifications in the device design and light-blocking configurations was explored. METHODS: A KPro was mounted on a contact lens holder on a bench, and forward light scatter was recorded with a camera attached to a rotating goniometer arm. Scattered light was recorded at different angles for different KPro modifications, and the point-spread function (PSF) curves were recorded. The area under the curve (AUC) was calculated for each PSF curve. RESULTS: The isolated KPro optical cylinder in a totally blackened holding lens had a tight PSF (AUC = 3.3). Additional blackening of the walls of the KPro stem did not further diminish forward scatter significantly. If the holding lens is made translucent by sandblasting (to simulate an in vivo carrier cornea) and the KPro is inserted without a backplate, forward scatter is substantial (AUC = 11.3). If a standard backplate (with holes) is added, light scatter is considerably reduced regardless of whether the backplate is made of polymethyl methacrylate or titanium (AUC = 5.3 and 4.4, respectively). Addition of an acrylic intraocular lens behind the KPro (the pseudophakic KPro setup) did not increase scatter. CONCLUSIONS: Most of the scattered light in eyes implanted with a KPro originates from the surrounding hazy corneal graft. The standard addition of a backplate reduces light scatter. There was no difference in forward light scatter between the aphakic and the pseudophakic KPro.