We use analogue experimentation to test the hypothesis that host rock competence primarily determines the morphology of kimberlite pipes. Natural occurrences of kimberlite pipes are subdivided into three classes: class 1 pipes are steep-sided diatremes emplaced into crystalline rock; class 2 pipes have a wide, shallow crater emplaced into sedimentary rock overlain by unconsolidated sediments; class 3 pipes comprise a steep-sided diatreme with a shallow-angled crater emplaced into competent crystalline rock overlain by unconsolidated sediments. We use different configurations of three analogue materials with varying cohesions to model the contrasting geological settings observed in nature. Pulses of compressed air, representing the energy of the gas-rich head of a kimberlitic magma, are used to disrupt the experimental substrate. In our experiments, the competence and configuration of the analogue materials control the excavation processes as well as the final shape of the analogue pipes: eruption through competent analogue strata results in steep-sided analogue pipes; eruption through weak analogue strata results in wide, shallow analogue pipes; eruption through intermediate strength analogue strata results in analogue pipes with a shallow crater and a steep-sided diatreme. These experimental results correspond with the shapes of natural kimberlite pipes, and demonstrate that variations in the lithology of the host rock are sufficient to generate classic kimberlite pipe shapes. These findings are consistent with models that ascribe the pipe morphologies of natural kimberlites to the competence of the host rocks in which they are emplaced.