Onset of function and consequent graft and patient survival in 7923 cadaveric kidney transplants (Tx) were analysed; 42.3% of grafts had early function, 43.6% delayed function, defined as a temporary need of dialysis postoperatively, and 14.1% grafts never functioned. Multivariate analysis of 1743 cases showed that important risk factors for delayed function were of non-immune origin, i.e. length of pre-Tx dialysis, or warm and cold ischaemia. The significant risk factors for non-functioning were of immune origin i.e. panel-reactive antibodies, DR mismatches, immunosuppression without cyclosporin A and previous Tx. Five-year survival rates of early function and delayed function grafts were identical when non-functioning grafts were excluded. Comparison of early function with delayed function grafts divided according to the length of the function delay showed worse (P < 0.05) survival in grafts with delayed function > 20 days only (50.6% grafts had delayed function < 10 days, 32.7% 10-20 days, and 16.7% > 20 days). Survival of recipients with non-functioning grafts was worse (P < 0.01) than those with early function and delayed function grafts. There was no difference in recipients' survival between early function and delayed function groups. When comparing early function and delayed function graft outcome, the problem is not so much a question of how as when to define early function. A statement that delayed function is prognostically a bad sign is not correct, as most delayed function grafts recover spontaneously without any effect on long-term graft and patient survival.(ABSTRACT TRUNCATED AT 250 WORDS)