The emergence of multidrug resistant Gram-negative pathogens, particularly carbapenemase producers, has forced clinicians to use last line antibiotics, such as colistin. Since colistin susceptibility testing presents several challenges, this study aimed at evaluating the performance of two alternative susceptibility methods for Klebsiella pneumoniae, namely, agar dilution (AD) and MIC test strips (MTS). These approaches were compared with the reference method, broth microdilution (BMD), and provide a quantitative description for the "skipped well" (SW) phenomenon. Colistin susceptibility was evaluated by BMD and AD in parallel and triplicate, using 141 K. pneumoniae clinical isolates while MTS performance was evaluated only for a subset (n = 121). Minimum inhibitory concentration analysis revealed that a substantial part (n = 26/141; 18.4%) of the initial isolates was deemed undetermined by BMD due to the following: discordance between replicates (1.4%); presence of multiple SWs (7.8%); and the combination of both events (9.2%). Both AD and MTS revealed a high number of false-susceptible strains ("very major errors"), 37.5% and 68.8%, respectively. However, AD agreement indices were reasonably high (EA = 71.3% and CA = 94.8%). For MTS these indices were lower, in particular EA (EA = 41.7% and CA = 89.6), but the approach enabled the detection of distinct sub-populations for four isolates. In conclusion, this study provides the most comprehensive study on the performance of AD and MTS for colistin susceptibility testing in K. pneumoniae, highlighting its limitations, and stressing the importance of sample size and composition. Further, this study highlights the impact of the SW phenomenon associated with the BMD method for K. pneumoniae.