Pores in the membrane of a bioartificial liver (BAL) allow it to function as a semipermeable barrier between its contents (i.e., liver cells) and components of the recipient's immune system. This study is designed to assess the influence of pore size on immune response to a BAL containing porcine hepatocytes. Sixteen healthy dogs were divided into four groups (four dogs per group) based on pore size of the BAL membrane and level of exposure to porcine hepatocytes. Group 1 dogs were administered porcine hepatocytes by intraperitoneal injection and served as positive controls. Group 2 dogs were exposed to porcine hepatocytes in a large-pore (200-nm) BAL, and group 3 dogs were exposed to porcine hepatocytes in a small-pore (10-nm) BAL. Group 4 dogs were exposed to a no-cell (unloaded) BAL and served as negative controls. Intraperitoneal injection of hepatocytes or 3 hours of BAL hemoperfusion was performed day 0 and 3 weeks later on day 21. Biochemical, humoral, and cellular measures of immune response were collected until day 44. The initiation of BAL hemoperfusion was associated with a rapid decline in CH(50) levels of complement and transient neutropenia and thrombocytopenia during all BAL exposures. Xenoreactive antibody response to BAL was increased by use of membranes with large pores and secondary exposures. Skin testing on day 42 showed a delayed-type hypersensitivity response to porcine hepatocytes that also correlated with level of previous antigen exposure. BAL treatment was associated with both immediate and elicited immunologic responses. The immediate response was transient and not influenced by membrane pore size, whereas elicited responses were influenced by pore size of the BAL during previous exposures.