Typical simple reaction time (RT) paradigms usually include a warning signal followed by a variable foreperiod before the presentation of a reaction stimulus. Most current interpretations suggest that the warning stimulus alerts the organism and so results in faster processing of either the sensory or motor components of the task. In this study, EMG was used to detect both covert and overt motor errors in a simple warned RT task. Results show that warning signals may trigger automatic motor activations which are likely to cause false alarms. Distribution analysis reveals that 77% of all errors detected with EMG are erroneous responses to the warning signal. Accordingly, we propose that movement triggering needs to be temporarily inhibited before the stimulus in order to prevent premature responses during the foreperiod. This proactive inhibition would be responsible for a paradoxical increase in RT for conditions with short foreperiods compared to control conditions in which no warning signal is presented. These results call for a reassessment of the theoretical framework used to interpret the effects of warning signals.