When designing cages to maximize welfare, we should consider both the physical and psychological needs of the animals for which the cage is intended. Many laboratory species show non-random defecation patterns and therefore might possess a psychological need for soiling sites with preferred characteristics. This study examined whether caged laboratory mice preferred to defecate on a floor with or without sawdust, and whether mice preferentially defecated in clearly partitioned areas of the cage which prevented mixing of soiled and clean sawdust. The mice were recorded as active in each area of the cage at a frequency that would be expected from the floor-area available, ie general activity was randomly distributed, but, the mice were selective in the areas used for defecation. Significantly more faeces were deposited in areas containing sawdust than in areas without (p<0.0001). After corrections for differences in the size of floor-area, significantly fewer faeces were deposited in partitioned areas at the rear of the cage than in the front area of the cage which contained the feeder and drinkers (p <0.001). Overall, these results show that the mice defecated in localized areas and preferred to defecate in areas containing sawdust. These results support other evidence which indicates that conventional cage designs for mice do not provide a sufficiently complex or appropriate environment to allow selective soiling behaviour. Such cages might therefore be inadequate with respect to catering for the psychological needs and overall welfare of laboratory mice.