The concentrations of the rare earth elements (REEs) in apatite from a 600-m-thick sequence toward the top of the Bushveld Complex, South Africa, have previously been published; they show two ranges of compositions in a lower section and an upper section. The data were interpreted in terms of liquid immiscibility. The evidence for this model is reassessed, and an alternative explanation offered that involves the trapped liquid shift effect. In this process, the true cumulus phase compositions reequilibrated with trapped liquid as it solidified, and REE abundances in apatite were variably increased. The extent of the change depends on the relative proportions of cumulus apatite and trapped liquid. Calculations show that if there was 5% apatite and 10% trapped liquid, as found in the lower section, the REE abundance in the final apatite would have only increased marginally relative to the original cumulus composition. However, in the upper section where there was only 2% apatite and 25% trapped liquid, the REE abundances were increased by a factor of 3. Most of the REEs are essentially incompatible in the other cumulus phases except apatite, and the other phases have little influence on the final concentrations of REEs in apatite. However, for Eu, which partitions into plagioclase, this mineral acts as a major buffering influence and relatively less Eu than the other REEs entered apatite during the solidification process. As a result, the calculated final apatite in the upper section developed a large negative Eu anomaly, as observed in the actual mineral analyses. The concept of large-scale liquid immiscibility in the upper part of the Bushveld Complex based on REEs in apatite is challenged, and an interpretation involving the trapped liquid shift effect is considered more plausible.