In order to describe the subjective (as opposed to the statistical) predictability of the retirement event, this study examined the timing of men's retirement relative to their own expectations. On-time and off-time retirements were assessed in a panel of older workers, participants in the VA Normative Aging Study, who were surveyed about work and retirement in 1978, 1981, and 1984. Over spans of two years, 66 percent of workers accurately predicted their eventual date of retirement (+/- 1 year), and 40 percent were exact to within three months. However, approximately one-third of workers did not accurately foresee their date of retirement, having either unfulfilled plans or unanticipated retirements. From a life course perspective, we interpret retirement as an orderly event when two-thirds of a sample can retire as planned. From a practical perspective (e.g., an interest in program development or research design), retirement is not orderly enough when one-third of exits are unscheduled.