Some theoretical considerations are given which show how a negative fluorescence dip (i.e., an increase in the monitored fluorescence signal) can be obtained when one is performing two-step excitation experiments in atmospheric pressure atomizers. Such an inverted dip is due to an increased population of the fluorescent level, caused by cascade de-excitation, both radiational and collisional, from the upper level(s) reached with the second excitation step. No coherence effects are involved. These considerations are supported by the experimental observation of the inverted dip in the stepwise, nonresonance fluorescence of silver atoms present in an inductively coupled argon plasma.