Fluorescent cell tracking dyes, in combination with flow and image cytometry, are powerful tools with which to study the interactions and fates of different cell types in vitro and in vivo.(1-5) Although there are literally thousands of publications using such dyes, some of the most commonly encountered cell tracking applications include monitoring of: stem and progenitor cell quiescence, proliferation and/or differentiation(6-8) antigen-driven membrane transfer(9) and/or precursor cell proliferation(3,4,10-18) and immune regulatory and effector cell function(1,18-21). Commercially available cell tracking dyes vary widely in their chemistries and fluorescence properties but the great majority fall into one of two classes based on their mechanism of cell labeling. "Membrane dyes", typified by PKH26, are highly lipophilic dyes that partition stably but non-covalently into cell membranes(1,2,11). "Protein dyes", typified by CFSE, are amino-reactive dyes that form stable covalent bonds with cell proteins(4,16,18). Each class has its own advantages and limitations. The key to their successful use, particularly in multicolor studies where multiple dyes are used to track different cell types, is therefore to understand the critical issues enabling optimal use of each class(2-4,16,18,24). The protocols included here highlight three common causes of poor or variable results when using cell-tracking dyes. These are: Failure to achieve bright, uniform, reproducible labeling. This is a necessary starting point for any cell tracking study but requires attention to different variables when using membrane dyes than when using protein dyes or equilibrium binding reagents such as antibodies. Suboptimal fluorochrome combinations and/or failure to include critical compensation controls. Tracking dye fluorescence is typically 10(2) - 10(3) times brighter than antibody fluorescence. It is therefore essential to verify that the presence of tracking dye does not compromise the ability to detect other probes being used. Failure to obtain a good fit with peak modeling software. Such software allows quantitative comparison of proliferative responses across different populations or stimuli based on precursor frequency or other metrics. Obtaining a good fit, however, requires exclusion of dead/dying cells that can distort dye dilution profiles and matching of the assumptions underlying the model with characteristics of the observed dye dilution profile. Examples given here illustrate how these variables can affect results when using membrane and/or protein dyes to monitor cell proliferation.