Previous studies have suggested that primary hepatocytes cultured in vitro undergo a rapid loss of function. On the other hand, in the clinic, drug induced liver injury typically manifests 5 days to 3 months after starting a medication. Thus, novel approaches that can maintain the function of primary human hepatocytes for longer durations of time may enable the development of improved in vitro assays for detecting hepatotoxicity. Previous studies have demonstrated that two-dimensional micro-patterning of hepatocytes with fibroblasts leads to improved maintenance of the hepatocyte phenotype relative to hepatocyte monocultures, in serum containing medium. Additionally, we, and others, have shown that three-dimensional culture of hepatocytes leads to enhanced function (in serum-free medium). In this study we wanted to (i) examine the effect of combining the above two approaches on hepatocyte function, and (ii) to further delineate the effect of serum on hepatocyte function. We developed a userfriendly and inexpensive approach for constructing layered spheroids. Similar to previous results in two-dimensional (2d) culture, we observed that 3d culture of hepatocytes alone (i.e. monoculture) in serum-containing medium led to an increase in the urea production rate, but near-complete loss of cytochrome activity in both lots of primary human hepatocytes (PHH) tested. In serum-free sandwich culture, cytochrome activity was maintained at the level observed in freshly thawed PHH for one lot, but almost completely lost in another lot. Spheroid culture of both lots of PHH in serum-free medium led to maintenance of CYP3A4 and CYP1A2 activity at the fresh thaw level, though CYP2B6 activity was reduced. In contrast to PHH monoculture, co-cultures of PHH with NIH 3T3 fibroblast cells benefitted from the presence of serum, and led to 3-5-fold increases in CYP activity relative to even serum-free spheroid monocultures. Layering of the fibroblasts did not result in improvements over mixed co-cultures. These results indicate the importance of appropriate serum-free monoculture control experiments in the evaluation of novel biomaterials and techniques for hepatocyte co-culture. Further, urea production and cytochrome production are decoupled; therefore, urea production is an insufficient readout when developing models for pharmaceutical applications.