An alternative theoretical explanation of integration of prior outcomes in risky decisions is the loss-sensitivity principle stating that a prior outcome is only added to expected losses. An experiment is reported which tested the implication of this principle that there will be less integration with an expected loss when its salience is reduced. A group of 20 undergraduates rated how likely they were to make imaginary roulette bets with less salient expected losses whereas another group with an equal number of undergraduates performed the same type of ratings for imaginary horse-race bets with more salient expected losses. In support of the implication, a prior outcome had a stronger impact in the latter group. A stronger impact was furthermore found for high as compared to low stakes. When the loss was salient, the impact of the prior outcome was greater for subjects who reported that they were in a positive mood than for subjects who reported that they were in a more neutral mood.