Analysis of proteomics data requires us to deal with many sources of variability, including both biological and technical variability. However, a further source of variability is introduced by the choice of analytical methods, where different decisions can be made regarding methods for normalisation, imputation, and tests for differential expression. In this talk, I will present our analysis of the most common methods for imputation, and their accuracy and performance on different kinds of data. I will then explore the impact of imputation method selection on the detection of differentially expressed proteins, both using established benchmarking datasets as well as experimental data. These results highlight the importance of decisions regarding analysis methods, and the collective impact such decisions can have on the identification of biologically meaningful variation in an experiment.