The efficacy of azathioprine in the treatment of multiple sclerosis was assessed by meta-analysis of the results of all published blind, randomised, controlled trials. 793 patients were enrolled in 5 double-blind and 2 single-blind studies. After 1 year of treatment, the increase in Kurtzke disability status score was no different in treated and control groups, but at 2 years there was a small difference (-0.22; 95% confidence interval [CI] -0.43, 0.003) in favour of azathioprine treatment; this difference was sustained, but not increased, after 3 years. The probability of freedom from any relapse during 1, 2, and 3 years' treatment was significantly greater in the azathioprine-treated group (relative odds over 3 years 1.97; 95% Cl 1.27, 3.04), but it is debatable whether the slight clinical benefits of azathioprine outweigh its side-effects.