The atmospheres of substellar objects contain clouds of oxides, iron, silicates and other refractory condensates. Water clouds are expected in the coolest objects. The opacity of these 'dust' clouds strongly affects both the atmospheric temperature-pressure profile and the emergent flux. Thus, any attempt to model the spectra of these atmospheres must incorporate a cloud model. However, the diversity of cloud models in atmospheric simulations is large and it is not always clear how the underlying physics of the various models compare. Likewise, the observational consequences of different modelling approaches can be masked by other model differences, making objective comparisons challenging. In order to clarify the current state of the modelling approaches, this paper compares five different cloud models in two sets of tests. Test case 1 tests the dust cloud models for a prescribed L-, L-T and T-dwarf atmospheric ( temperature T, pressure p, convective velocity v(conv)) structures. Test case 2 compares complete model atmosphere results for given ( effective temperature T-eff, surface gravity log g). All models agree on the global cloud structure but differ in opacity relevant details such as grain size, amount of dust, dust and gas-phase composition. These models can loosely be grouped into high- and low-altitude cloud models whereas the first appears generally redder in near-infrared colours than the latter. Comparisons of synthetic photometric fluxes translate into a modelling uncertainty in apparent magnitudes for our L-dwarf (T-dwarf) test case of 0.25 <= Delta m <= 0.875 (0.1 <= Delta m <= 1.375), taking into account the Two-Micron All Sky Survey, the UKIRT WFCAM, the Spitzer IRAC and VLT VISIR filters with UKIRT WFCAM being the most challenging for the models. Future developments will need closer links with laboratory astrophysics, and a consistent treatment of the cloud chemistry and turbulence.