A cone field apparatus was developed in which spatial working memory (WM) and reference memory (RM) can be assessed automatically and simultaneously. The cone field is a large open field with 16 cones in it. A configuration of four cones was baited with a food pellet. In the holeboard, the predecessor of the cone field, no clear criteria exist to distinguish between accidental and directed orientations toward the target (hole). In the cone field the visit to the target (cone) is unambiguously defined. We found that senescent rats were impaired on both the WM and the RM. The development of a fixed pattern of visiting the baited cones was effectively prevented by starting the rats randomly from four different positions. When only one start position was used, rats developed a fixed pattern of visiting the baited cones. This pattern was more pronounced in the young than in the senescent animals. In the last experiment we explored whether the task might be suited for the assessment of WM and RM performance in longitudinal studies. Rats performed like naive animals on both memory components when re-acquisitions were separated by 4-month intervals without training.