Mineral inclusion assemblages in zircon are almost certainly a complex function of crystallization sequence, magma chemistry, and physical proximity to other crystallizing phases. While the latter may be largely random, the former two factors should have more systematic controls. To better constrain the effects of crystallization sequence, we investigated a suite of granitoids from the Cretaceous arc of southern California and from the Miocene Colorado River Extensional Corridor, ranging from tonalites to leucogranites. In particular we examined inclusions in a variety of accessory minerals to elucidate crystallization order effects. Apatite is almost universally overrepresented as an inclusion in the studied minerals relative to its volumetric abundance in the rock as a whole. Where crystallization order can be inferred, earlier phases usually either have a higher proportion of apatite inclusions (11 out of 20 cases) or show no significant differences in apatite content from later phases (7 cases). An increase in the proportion of apatite with progressive crystallization is observed in four out of the 20 samples with clear crystallization orders (two rocks out of the 20 contain both an increasing and a decreasing case among different mineral pairs and are included in both groups). We interpret these observations as showing the widespread early crystallization of apatite inclusions – likely by pileup of the slow-diffusing phosphorus complex along growing mineral grains. The lower apatite contents in late-crystallizing modal phases are also reflected in the very low apatite contents within late-crystallizing zircon from some studied mafic-intermediate units, contrary to earlier suggestions that apatite decreases as a proportion of the inclusion assemblage with increasing whole rock silica. Plagioclase inclusions in zircon are on average more albitic than matrix plagioclase. In some samples, sequences of plagioclase crystallization can be identified by albite content among 1) magmatic cores of matrix grains, 2) inclusions in mafic accessory minerals, 3) inclusions in zircon, and 4) post-magmatic alteration veins and exsolution lamellae. However, the relative proportions of quartz, K-feldspar, and plagioclase inclusions do not systematically vary in accessory minerals with crystallization order, suggesting other origins of the often-recognized modal shifts toward more quartz- and K-feldspar-rich inclusion assemblage compared to the matrix. Unlike Himalayan leucogranites, when muscovite is present in the matrix in our sample suite, it is a rare inclusion phase in later-crystallizing accessories and may, in some cases, even be absent in zircon even when present in other phases. It is possible to mistake non-magmatic inclusions in detrital grains for primary magmatic inclusions, and examination of internal host grain texture and geochronology (when applicable) is especially important in a detrital setting. These results should allow an improved framework for interpreting inclusion assemblages in detrital magmatic minerals and in interpreting the phase petrology of the host igneous rock. Given the remaining ambiguities in linking a number of observations (i.e., apatite contents, position on the quartz-alkali feldspar-plagioclase ternary, and the ratio of felsic to mafic silicate phases) to specific magmatic provenance, the identification of rare but petrologically significant inclusion phases such as muscovite, halides, Mn and Nb minerals, sulfides, or baddeleyite, may be more helpful in pointing to either highly evolved or more mafic provenances.