Synthetic lipid membranes in the absence of proteins can display quantized conduction events for ions that are virtually indistinguishable from those of protein channel. By indistinguishable we mean that one cannot decide based on the current trace alone whether conductance events originate from a membrane which does or does not contain channel proteins. Additional evidence is required to distinguish between the two cases, and it is not always certain that such evidence can be provided. The phenomenological similarities are striking and span a wide range of phenomena: The typical conductances are of equal order and both lifetime distributions and current histograms are similar. One finds conduction bursts, flickering, and multistep-conductance. Lipid channels can be gated by voltage, and can be blocked by drugs. They respond to changes in lateral membrane tension and temperature. Thus, they behave like voltage-gated, temperature-gated and mechano-sensitive protein channels, or like receptors. Lipid channels are remarkably under-appreciated. However, the similarity between lipid and protein channels poses an eminent problem for the interpretation of protein channel data. For instance, the Hodgkin-Huxley theory for nerve pulse conduction requires a selective mechanism for the conduction of sodium and potassium ions. To this end, the lipid membrane must act both as a capacitor and as an insulator. Non-selective ion conductance by mechanisms other than the gated protein-channels challenges the proposed mechanism for pulse propagation. Nevertheless, the properties of the lipid membrane surrounding the proteins hardly ever enter into the textbook discussion of membrane models. Some important questions arise: Are lipid and protein channels similar due a common mechanism, or are these similarities fortuitous? Is it possible that both phenomena are different aspects of the same phenomenon? Are lipid and protein channels different at all? In this review we will document some of the experimental and theoretical findings that show the similarity between lipid and protein channels. We discuss important cases where protein channel function is strongly correlated to lipid properties. Based on some statistical thermodynamics simulations we discuss how such a correlations could come about. We suggest that proteins can in principle act as catalysts for lipid channel formation and that some apparently mysterious correlations between protein and lipid membrane function can be understood in this manner.