Ice crystals in cirrus clouds can orient themselves depending on their size, shape, and the hydrodynamical conditions. Ice crystal orientation affects the radiative impact of clouds and can introduce biases in retrievals of cloud properties. Horizontally oriented ice crystals such as plates, for example, specularly reflect sunlight and produce the so-called cloudglint. This paper introduces a parameterization of horizontally oriented ice crystals for Monte Carlo radiative transfer simulations, which is based on the treatment of reflection by a statistically rough ocean surface. The parameterization includes polarization and allows for comparably fast simulations of the cloudglint without having to store large, multidimensional tables of scattering phase functions. Oriented ice crystals are approximated by plates with infinitely extended top and base faces without sides and a Gaussian orientation angle distribution. The parameterization includes external reflections as well as an internal reflection and transmission through the ice crystal. Multiple scattering between different ice crystals is accounted for. The parameterization was validated by a comparison with cloudglint simulations of the raytracing algorithm CrystalTrace. Deviations between CrystalTrace and the parameterization for large solar zenith angles and large aspect ratios can be explained by the influence of the crystal sides, which are neglected in the parameterization. Possible applications of the parameterization include, for example, the detection of even small amounts of (oriented) ice crystals from polarized observations of the cloudglint or the improvement of retrievals of cloud properties.