Three experiments were designed to examine the processing of nominal information during a visual search. In Experiment I subjects searched for a single pre-specified target letter, and the stimulus exposure-time needed to yield a 50% (corrected for chance) level of accuracy was estimated using the PEST procedure. The results showed that the exposure-time was not influenced by the presence (in the irrelevant items) of the target's other case, and this suggested that there was no obligatory accessing of nominal information during preattentive (Neisser, 1967) processing. In Experiment n subjects searched in one condition for a single target which was specified as being one of a same-name pair of letters, and in a second condition for a target specified as one of a same-shape pair of letters. The exposure-time required for the same-name search was greater than that required for the same-shape search, and this suggested that even when nominal information might be expected to aid performance, it was either not accessed during the search, or if accessed, was relatively ineffective as a basis for selection. In Experiment m accuracy of search was compared in the sameshape and same-name conditions with a control condition. The target in this control condition was specified as one of a pair of letters not sharing a name or any special shape. The results confirmed the difference detected in Experiment 11, but they did not show any difference between performance in the same-name condition and in the control condition. This last finding indicated that subjects were unable to access the name of the target during the first stage of analysis. Consequently, it was suggested that some focal processing is required for the development of a sufficiently detailed level of visual representation to allow for naming.