The accuracy of disk susceptibility methods for colistin against 778 bacterial pathogens was evaluated in comparison with Etest using interpretive criteria available from the Clinical and Laboratory Standards Institute (CLSI). Colistin exhibited excellent activity against Acinetobacter baumannii and Escherichia coli isolates (minimum inhibitory concentration for 90% of the organisms (MIC(90))=0.5 mg/L), whilst it was less active both against Enterobacter spp. and Klebsiella pneumoniae (MIC for 50% of the organisms (MIC(50))=0.5 mg/L, MIC(90)=16 mg/L). Colistin also showed good activity against Pseudomonas aeruginosa (MIC(90)=2 mg/L, MIC(50)=1 mg/L) but poor activity against Stenotrophomonas maltophilia (MIC(50)=8 mg/L, MIC(90)=128 mg/L). Only 0.8% of minor errors were observed between the studied methods for P. aeruginosa isolates when the CLSI criteria were applied. All A. baumannii isolates with a zone diameter < or =12 mm were resistant and those with a zone diameter > or =14 mm were susceptible according to MIC breakpoints established by the CLSI. Among nine isolates exhibiting a zone diameter of 13 mm, one was resistant to colistin (MIC=8 mg/L) and eight isolates were susceptible (MIC=0.5 mg/L). Applying a MIC breakpoint of < or =2 mg/L for susceptibility in Enterobacteriaceae, all isolates with a zone diameter > or =14 mm were susceptible, whilst all isolates with a zone diameter < or =11 mm were resistant. Among isolates with zone diameters of 12-13 mm, 59% were characterised as susceptible. Major errors were observed only in K. pneumoniae isolates at a rate of 0.8%. The poor agar diffusion characteristics of colistin limit the predictive accuracy of the disk diffusion test and consequently values of 12-13 mm should be confirmed with MIC determination by Etest or broth dilution method.