To determine whether concealed imagery was an effective mnemonic aid in the recall of paired objects, two studies were conducted with explicitly worded instructions to conceal targets and with variable image formation periods. In each study, 40 subjects were presented with counterbalanced verbal descriptions of Concealed, Pictorial, and Separate object interactions; the descriptions were further categorized according to the type of image elicited, whether Normal or Bizarre. Results of Experiment 1 showed that speed demands forced subjects to form concealed images rapidly, with a subsequent detrimental effect on their recall, particularly in the Bizarre Concealed condition. When more time was allowed for image construction and processing, the pattern of findings changed dramatically in Experiment 2: Bizarre Concealed images were recalled just as well as Pictorial ones. In support of N. H. Kerr and U. Neisser (1983), the mnemonic effectiveness of these images is determined by their formation times rather than by instructions to conceal. (Four tables of data are included, and 22 references are attached.) (SR)