All 134 episodes of bacteremia caused solely by Pseudomonas aeruginosa in a university hospital in the periods 1976-1982 and 1992-1996 were reviewed retrospectively to determine the clinical manifestations, outcome and prognostic factors. The mortality for the 30-day interval after drawing the first positive blood culture was 41%, but dropped from 53% in the first period to 29% in the second period (P=0.006). Mortality was highest in patients treated with an aminoglycoside only, as against those treated with other appropriate antibiotics (55% versus 25%, P=0.001). Over the two decades studied, use of an aminoglycoside only decreased, use of paracetamol (=acetaminophen) increased, and removal of both urinary and blood vessel catheters became more common. The mortality was 18% in patients with catheter removal (46% in the other patients, P=0.017) and 27% in patients who received paracetamol around the time of drawing the first positive blood culture (50% for the other patients, P=0.010). Logistic regression analysis showed that shock, central nervous system involvement, preceding thromboembolism and rapidly fatal underlying disease were associated with a fatal outcome, whereas catheter removal, appropriate antibiotic therapy and paracetamol therapy were associated with survival. The improved prognosis of Pseudomonas aeruginosa bacteremia over the two decades is thus due mainly to three changes in management of the infection: the more frequent use of new anti-pseudomonal beta-lactams and ciprofloxacin instead of aminoglycosides as monotherapy; the more frequent practice of removing catheters; and the increased use of paracetamol around the time of drawing the first positive blood sample.