The form of S's silent verbal encodings of briefly presented stimuli was manipulated. Order of encoding accounted for the effects of instructions to attend selectively on accuracy of report. Instructions to attend selectively had no significant effect on Ss who encoded the stimulus in the order dictated by English syntax. The Ss whose coding strategy permitted them to encode first the stimulus attribute they were told to attend to reported this attribute more accurately than incidental attributes. The superior accuracy was not a by-product of order of report. The attribute encoded 1st is based on a better visual trace and also may be retained better in short-term memory. (17 ref.) (PsycINFO Database Record (c) 2006 APA, all rights reserved)