In this study, the performance of phase separated and crystallized amorphous solid dispersions (ASDs) was evaluated by non-sink in vitro dissolution testing in fasted-state simulated intestinal fluid (FaSSIF) and in vivo in rats. The amorphous phase-separated or crystallized ASDs were prepared by mixing an ASD of the model drug celecoxib (CCX) in polyvinylpyrrolidone (PVP) with pure amorphous or micronized crystalline CCX at 20, 40, 60 or 100% of the total drug load (25:75 w/w CCX:PVP), respectively. As expected, crystallization of CCX in the ASDs generally had a negative influence on both the area under the curve of the dissolution curve (in vitro AUC) and the plasma concentration-time profile (in vivo AUC) in rats compared to the pure ASD. However, the difference between the in vivo AUC of the pure ASD and the 20% and 40% crystallized ASDs was not statistically significant, which could indicate that a low fraction of crystallization of a drug in an ASD may only have limited impact on in vivo performance and hence bioavailability. In comparison, amorphous phase separation of CCX in the ASDs did not negatively influence the in vitro AUC and in vivo AUC to the same degree as crystallization and the dissolution profiles of all the amorphous phase-separated ASDs were similar to that of the pure ASD. In fact, even though a slight decrease of in vivo AUC with increasing fraction of amorphous phase separation was observed, the 20% and 40% amorphous phase-separated ASDs were bioequivalent with the pure ASD.