College undergraduates participated in a cost/benefit analysis and a speed/accuracy analysis of semantic priming in a lexical decision task. In both studies, half the cues were neutral and half were words from 30 semantic categories. Word targets were the category names, and nonword targets were derived from those names. The cue-word was valid 80% of the time and invalid 20% of the time. In the first experiment, cue time, a between groups factor, was either 200, 300, 400, 500, or 700 milliseconds (msec). It was found that valid cues produced facilitation at cue times as short as 400 msec, and invalid cues produced inhibition at cue times as short as 200 msec. The second experiment used a response-signal technique to collect information about the speed/accuracy trade-off in a lexical decision task. Cues were always presented for 800 msec, and targets were presented for varying time periods. Results indicated that subjects are capable of trading accuracy for speed, invalid cues can lower discriminability, and response bias in an active component in priming. The two experiments indicate that discriminability, criterion bias, and response bias seem to be integrated in providing facilitation, but that simple applications of criterion bias or response bias alone do not adequately explain the facilitation effect. (Author/GT)