The impact of shorter hospital lengths of stay on patient outcomes at discharge from acute care after knee arthroplasty was investigated in a prospective observational outcome study at three Melbourne public hospitals during a 5-month period from October 1999 to March 2000. The participants were 105 consecutive patients (35 at each hospital), with a mean age of 71 years. Outcome measures were length of stay, destination (home or rehabilitation) and functional mobility at discharge from the acute care facility. During the study period mean hospital length of stay across the three hospitals was 6.5 days, more than 30% less than the Victorian average for the preceding year. This was associated with high rates of discharge to rehabilitation facilities (mean 64%), with rates varying between the three hospitals (97%, 57% and 40%). However, in each hospital, one-third of this group had already achieved a level of independent functional mobility adequate for discharge home, highlighting an apparent influence of non-clinical factors on discharge decisions, including pressure to decrease length of stay, hospital policy and availability of a rehabilitation bed. Ways of achieving discharge directly home for a greater number of patients following knee arthroplasty and of determining optimal length of stay are discussed.