The ability to identify and recognize objects in space is essential in the day-to-day life of most species. This process, however, is seldom restricted to the visual domain. Sensory input from senses other than vision, can aid the recognition of objects or change their contextual meaning. For example, when we feel and hear that it’s raining, we might use vision to search for a place to take shelter. To investigate the neural underpinning of this behavior we developed a task for head-fixed mice in virtual reality which can be easily combined with large-scale neuronal recordings. Contrary to most published virtual reality tasks, which consist of corridors of monochrome gratings and patterns, we used a gaming engine (Unity) to create a naturalistic looking virtual environment. Mice run on a treadmill to traverse a straight path in the environment where three different objects are placed on semi-random locations. When mice enter the environment, an auditory contextual cue starts playing (birdsong or rain), mice can obtain a reward at object A or object B, dictated by the auditory cue. Object C, however, is never rewarded. Mice typically fully acquire the paradigm in six weeks and are motivated to run a high number of trials (> 200), even with moderate food restriction. We expect this paradigm to be useful for researching the neural basis of multisensory contextual object recognition in mice.