This study analyzed the early effects of the Medicare Prospective Payment System (PPS) on the likelihood of hospital's discharging Medicare beneficiaries to skilled nursing facilities (SNFs), intermediate care facilities (ICFs), and home health agencies. It also examined length of stay before transfer. Discharge abstract data on patients in five DRG groups were studied. Data were obtained from 501 hospitals for the third quarters of 1980, 1983, 1984, and 1985. Multinomial logit and ordinary least squares regression techniques were employed. After controlling for hospital and patient characteristics, including severity of illness, it was found that the probability of transfer increased substantially in virtually all DRGs and discharge destinations studied. This was particularly true for patients with stroke, pneumonia, and major joint and hip procedure. The analysis reveals that PPS increased the rate of discharges to subacute facilities. This effect was stronger for transfer to SNFs than to ICFs and home health agencies. Further, the impact of PPS on transfers was greater in 1985 than in 1984. Lengths of stay before transfer tended to decline in almost all DRGs and destinations examined. However, the effects of PPS on lengths of stay of transferred patients were not statistically significant at conventional levels. The results suggest that payment experiments with broader forms of bundled services are in order, as are experiments with hospital acute-subacute swing beds.