With current single-cell proteomics approaches, we can confidently detect the proteome of larger single cells, such as frog or fish embryos. Although still very difficult, smaller cells such as mammalian cells can also be detected by various methods. However, certain limitations such as low throughput, tedious sample handling and difficulties in quantification severely limit the use of single-cell proteomics on large-scale studies. Moreover, there are also many important cell species that are far smaller than embryos or mammalian cells. Current instrumentations and methods are simply not sensitive enough to detect the proteome from a single mitochondrion or a single bacterium. Here we present a novel super-resolution proteomics method to differentiate sub-populations of cells at single-cell resolution. With the combination of newly developed chemical derivitization, advanced statistical modeling, and optimized sample preparation process, we have shown that sub-populations of cells could be differentiated at the single-cell level with a typical throughput of >2,000 cells per day and >1,500 proteins quantified per sample.