Downstream-bioprocess operations, for example, selective flocculation, are inherently variable due to fluctuations in feed material, equipment performance, and quality of additives such as flocculating agents. Due to these fluctuations in operating conditions, some form of process control is essential for reproducible and satisfactory process performance and hence, product quality. Both product (alcohol dehydrogenase) and key contaminants (RNA, protein, cell debris) within a Saccharomyces cerevisiae system were monitored in real-time adopting an at-line enzymatic reaction and rapid UV-VIS spectral-analysis technique every 135 seconds. The realtime measurements were implemented within two control configurations to regulate the batch-flocculation process according to prespecified control objectives, using the flocculant dose as the sole manipulative variable. An adaptive, model-based control arrangement was studied, which combined the rapid measurements with a process model and two model parameter-identification techniques for real-time prediction of process behavior. Based on an up-to-date mathematical description of the flocculation system, process optimization was attained and subsequent feedback control to this optimum operating set point was reproducibly demonstrated with a 92% accuracy. A simpler control configuration was also investigated adopting the cell debris concentration as the control variable. Both control arrangements resulted in superior flocculation-process performances in terms of contaminant removal, product recovery, and excess flocculant usage compared to an uncontrolled system.