Mixing effects in protein precipitation processes are very poorly understood. They impact apparent protein solubility, protein structure in precipitates, particle size, morphology and recoverable activity. In this work, a model is proposed to describe the impact of mixing on protein solubility in precipitation processes. A series of semibatch-mode precipitation experiments were performed using bovine liver catalase and ammonium sulfate to test model predictions. The process variables studied include initial protein supersaturation, agitation rate, and rate of addition of salt solution to the initial protein solution. The mixing model parameters were estimated independently with the aid of a series-parallel dye reaction. The results showed good agreement between model and experiment. The initial protein supersaturation was the most important variable at the 1-L scale. The influence of mixing on recoverable activity was also examined. Scale-up guidelines are suggested based on model simulations and experimental results.