The first step in the implementation of protein crystallization experiments by batch and vapor diffusion techniques is the mixing of the solution of the protein and the solution of the precipitating agent to form a drop. It is demonstrated by optical microscopy and by Mach−Zhender interferometry that, depending on the relative values of the properties of both solutions and on the sequence of mixing (protein over precipitant solution or vice versa), complex but well-defined fluid patterns occur. The formation of these patterns is governed by phenomena such as density-driven flow, Rayleigh−Taylor instability, diffusion reaction, and double-diffusive instabilities (both salt fingering and layering). The formation of local supersaturation maxima and minima in the crystallization drop due to these phenomena explains why and how the protocol of mixing may affect the crystallization results. As a practical application derived from this study, it is recommended that one perform active mixing when seeking reproducibility and avoid active mixing when seeking a more complex and extensive screening of the crystallization space.