The ability of 1-year-old infants to remember the location of a nonvisible target was investigated in 3 experiments. Infants searched for a toy hidden in one of many possible locations within a circular bounded space. The presence, number, and spatial arrangement of local cues or "landmarks" within this space were varied. The results of Experiment 1 showed that search performance was highly successful when a landmark was coincident with the location of the toy ("direct"), but less successful when a landmark was adjacent to the target location ("indirect"). The results of Experiment 2 suggested that search with an indirect landmark may be more fragile than search with no landmarks at all. In Experiments 3a and 3b, 2 different configurations of indirect landmarks were employed; search performance was equally poor with both of these and was inferior to search with no landmarks. It is concluded that infants of this age are able to associate a nonvisible target with a direct landmark and are able to code the distance and direction of a target with respect to themselves or with respect to the larger framework. However, there was no evidence that they can code the distance and direction of a target relative to another object. The difficulty of coding with indirect landmarks is interpreted in terms of cognitive complexity and conflict between spatial strategies.