After a failure to obtain a recency effect in the perception of ambiguous figures three experiments were carried out to determine the conditions which govern the effect. It was found that the duration of the stimulus which precedes the ambiguous figure (the most recent stimulus) was important. Also of importance was the interval between the next-to-most-recent stimulus and the ambiguous figure. The effect of manipulating these variables was such that sometimes no recency effect was obtained, implying that their state is sometimes critical for the effect. Variations in the duration of the ambiguous stimulus did not cause the recency effect to vary. Recommendations are made for the demonstration of a reliable recency effect.