Susceptibility testing for colistin remains challenging primarily due to its inherent properties. We evaluated colistin stability in agar and reproducibility of colistin MICs obtained by agar dilution, broth macro- and micro-dilution and MIC gradient strips on 3-7 iterations of each method using clinical Klebsiella pneumoniae (susceptible-CS, and resistant-CR, n = 2 each), mcr-harboring Escherichia coli (n = 2), and reference strains E. coli ATCC25922 and Pseudomonas aeruginosa ATCC27853. MICs for reference strains were not in the given range using Etest and broth microdilution (ATCC25922, 0.125 and 4 μg/ml, respectively). MICs of CR-1 and CR-2, and of the mcr-harboring E. coli showed high concordance between agar and broth dilution varying up to one 2-fold dilution. However, remarkable variations were observed on broth dilution with CS-1 and CS-2 (MIC range 0.25-32 and 0.5-64 μg/ml, respectively); whereas for agar dilution the MIC for both CS strains was 0.5 μg/ml in all the runs. MICs obtained by MIC gradient strips were lower than those obtained by dilution methods (1-2 dilutions for CS and mcr strains, and up to five dilutions for CR strains). To confirm uniform distribution of colistin in agar, a single strain was spotted in five different regions of the same plate. All spots showed concordant growth with maximum one dilution difference. No effect on MIC was found due to storage of colistin-containing agar plates for 7 days at 4 °C. In our hands, agar dilution was superior in terms of reproducibility and robustness, compared to broth dilution methods, for colistin MIC determination.