We propose a new masking technique for masking word stimuli. Drawing on the phenomena of metacontrast and paracontrast, we alternately presented two prime displays of the same word with the background color in one display matching the font color in the other display and vice versa. The sequence of twenty alterations was sandwich-masked by structure masks. Using this masking technique, we conducted evaluative priming experiments with positive and negative target and prime words. Significant priming effects were found – for primes (...) and targets drawn from the same as well as from different word sets. Priming effects were independent of prime discrimination performance in direct tests and they were still significant after the sample was restricted to those participants who showed random responding in the direct test. (shrink)