Given that laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) has the potential to directly determine the concentrations of some 80 elements in solid samples, the fact that most applications are limited to considerably smaller numbers of analytes is indicative of the inherent problems with calibration. These stem from elemental response variations, both between analytes in any given sample and between matrices for any given analyte. Although response variations are often attributed to differences in ablation or transport efficiencies, there have also been indications that some degree of elemental fractionation may occur in the ICP. The results of the present investigations demonstrate that the ICP is the predominant source of fractionation, and thus response variations are related to the thermodynamic properties of the elements and their host particles. By studying analyte response as a function of carrier gas flow rate (so called flow rate plots) in 16 matrices, patterns in the behaviour of the elements in LA-ICP-MS could be clearly discerned and used for classification. Three groups of elements displaying consistent behavioural patterns over all matrices were identified from these studies: Group A, comprising refractory elements with high oxide bond dissociation enthalpies; Group B, including the rare and heavier alkaline earth (Ca, Sr, Ba) elements; and Group C, consisting of volatile or low mass elements. As each group exhibits decidedly different optimum flow rates and flow rate plot shapes which, with the exception of the group C elements, also depend on the matrix, the utility of LA-ICP-MS for multi-element analyses is severely compromised. In fact, quantitative determination of a wide range of analytes demands that calibration factors be established for at least one element from each group, as well as for all elements that could not be satisfactorily classified. This classification may serve as a guide in the selection of suitable internal standards for LA-ICP-MS, at least for certain groups of analytes. Examples are also given showing how flow rate plots can be employed to predict the adequacy of selected internal standards or solid standard materials for calibration.