One hundred and fifty-four patients with histologically verified non-alcoholic chronic liver disease were randomized to azathioprine or prednisone treatment. After a median of 91 months observation time, the cause of death was assessed retrospectively. Autopsy was performed in 82% of 71 deaths. In the azathioprine group 33% (13/39) died from malignant neoplasia, and in the prednisone group (13%) (4/32) (p = 0.08). Considering a possible fatal outcome as a consequence of treatment, this finding urges caution in the long-term application of azathioprine at the usual dose level.