Scheelite (CaWO 4) is a widespread accessory mineral associated with hydrothermal veins in the Archean gold deposits of the Kalgoorlie-Norseman region of Western Australia. Rare earth element (REE) and other trace element abundances in scheelite have been determined in situ by excimer laser ablation-inductively coupled plasma-mass spectrometry (ELA-ICP-MS) in order to constrain the composition and sources of the mineralizing fluids. The scheelites can be grouped according to the two distinct types of chondrite-normalized REE (REE(N)) patterns which they exhibit: hump-shaped (type I) and flat (type II). Type I scheelites have higher total and Na concentrations than type II samples and can be further subdivided into type Ia which has maximum REE(N) concentrations between Sm and Gd, and type Ib which has maximum REE(N) concentrations displaced toward Dy. Type I scheelites are dominant at Coolgardie, Kalgoorlie, and Kambalda, whereas type II samples are most abundant at Norseman. A few scheelite grains exhibit both type I and type II characteristics. The two types of REE(N) patterns with different Na abundances suggest that the trivalent REE substituted into the Ca site of scheelite by two different mechanisms: 2Ca 2+ = REE 3+ + Na + for the type I scheelites and 3Ca 2+ = 2REE 3+ + [ ](Ca) (where [ ](Ca) is a Ca site vacancy) for the type II samples. The parabolic shape of REE(N) patterns in the type I scheelites can be used to calculate the ionic radius of the REE 3+ that preferentially substitutes into the scheelite structure. When type I REE patterns are normalized to account for variations in light REE abundance, the average value obtained is 1.060 Å for type Ia and 1.055 Å for type Ib scheelites. These values are smaller than the value of 1.12 Å for the Ca site but are in excellent agreement with the value of 1.06 Å predicted by the coupled Na +-REE 3+ substitution mechanism, assuming size as well as charge compensation. This result implies that Na + and REE 3+ substitute into adjacent sites in the crystal structure of type I scheelites. In contrast, the incorporation of REE into type II scheelites appears to be independent of size. It is suggested that in this type of scheelite, involvement of a Ca site vacancy relaxes the size constraint on the substitution of the REE 3+. The influence of REE speciation in the hydrothermal fluid on the systematic difference between scheelite-fluid partition coefficients for type I and II scheelites is of secondary importance compared with crystal chemistry effects. The distinctive REE(N) patterns and high Na contents of type I scheelites imply that they crystallized from hydrothermal fluids with higher Na activities than those that formed type II samples. Type I scheelites exhibit no changes in the size of the Eu anomaly with REE concentration, implying a predominance of Eu 3+ and crystallization under relatively oxidized conditions. Type II scheelites have variable Eu anomalies and trivalent REE concentrations and thus appear to contain mostly Eu 2+ and to have formed under reduced conditions. Type Ia and II scheelites have (Ce/Lu)(N) > 1 and are interpreted to have crystallized from light REE-enriched fluids, whereas type Ib scheelites with (Ce/Ln)(N) < 1 formed from light REE-depleted fluids. The fluid characteristics inferred from the REE and trace element systematics in the scheelites can be attributed to the rock types and estimated physical conditions in the Kalgoorlie-Norseman region at the time of hydrothermal activity. High Na activities are consistent with buffering of fluids by greenschist facies mafic rocks, whereas low fluid Na activities are a product of fluid-rock reactions at higher metamorphic grades. Oxidized fluids are consistent with near-surface derivation, whereas reduced fluids are indicative of generation at elevated temperatures and pressures. The range and distribution of fluid (Ce/Ln)(N) ratios are similar to those of the major rock types in the region.