This paper presents a re-analysis of previously reported data in which the mean correlation over 10 experiments between vividness of visual imagery for faces and facial recognition memory was .247. When task accuracy was consistently scored to eliminate false-alarms, the correlation dropped to .145, although it remained above zero. Since the vividness/task confidence correlation of .176 was also significant, it is suggested that the weak vividness/accuracy relationship may reflect different processing requirements in the vividness and recognition tasks and subjectivity of the vividness rating scale.