The transport of colloid-contaminant complexes and colloid-sized pathogens through soil to groundwater is of concern. Visualization and quantification of pore-scale colloid behavior will enable better description and simulation of retention mechanisms at individual surfaces, in contrast to breakthrough curves which only provide an integrated signal. We tested two procedures for quantifying colloid movement and retention as observed in pore-scale image sequences. After initial testing with static images, three series of images of synthetic microbead suspensions passing through unsaturated sand were examined. The region procedure (implemented in ImageJ) and the Boolean procedure (implemented in KS400) yielded nearly identical results for initial test images and for total colloid-covered areas in three image series. Because of electronic noise resulting in pixel-level brightness fluctuations the Boolean procedure tended to underestimate attached colloid counts and conversely overestimate mobile colloid counts. The region procedure had a smaller overestimation error of attached colloids. Reliable quantification of colloid retention at pore scale can be used to improve current understanding on the transport mechanisms of colloids in unsaturated porous media. For example, attachment counts at individual air/water meniscus/solid interface were well described by Langmuir isotherms.