Most studies of sandstone provenance involve modal analysis of framework grains using techniques that exclude the fine-grained breakdown products of labile mineral grains and rock fragments, usually termed secondary matrix or pseudomatrix. However, the data presented here demonstrate that, when the proportion of pseudomatrix in a sandstone exceeds 10%, standard petrographic analysis can lead to incorrect provenance interpretation. Petrographic schemes for provenance analysis such as QFL and QFR should not therefore be applied to sandstones containing more than 10% secondary matrix. Pseudomatrix is commonly abundant in sandstones, and this is therefore a problem for provenance analysis. The difficulty can be alleviated by the use of whole-rock chemistry in addition to petrographic analysis. Combination of chemical and point-count data permits the construction of normative compositions that approximate original framework grain compositions. Provenance analysis is also complicated in many cases by fundamental compositional alteration during weathering and transport. Many sandstones, particularly shallow marine deposits, have undergone vigorous reworking, which may destroy unstable mineral grains and rock fragments. In such cases it may not be possible to retrieve provenance information by either petrographic or chemical means. Because of this, pseudomatrix-rich sandstones should be routinely included in chemical-petrological provenance analysis. Because of the many factors, both pre- and post-depositional, that operate to increase the compositional maturity of sandstones, petrologic studies must include a complete inventory of matrix proportions, grain size and sorting parameters, and an assessment of depositional setting.