According to evolutionary theory, emotions are psychological mechanisms that have evolved to enhance fitness in specific situations by motivating appropriate (adaptive) behavior. Taking this perspective, a previous study examined the relationship between mood and preference for natural environments. It reported that participants' anxiety level was associated with a preference for landscapes offering what Appleton called "refuge," while participants' anger and cheerfulness were both associated with a preference for landscapes offering what Appleton called "prospect." We attempted to replicate these results and to improve on the study by experimentally manipulating mood. Using a between-subjects design, 80 participants were instructed to self-induce one of four moods: anger, sadness, anxiety, or joy. After the mood induction, they viewed fourteen landscape photographs and recorded the seven most preferred. It was hypothesized that subjects experiencing anger or joy would prefer landscapes rich in "prospect" features, whereas participants experiencing sadness or anxiety would prefer landscapes rich in "refuge" features. In contrast to the previous study, the predictions were not supported: artificially induced moods may not provide ecological validity as a test of the "mood as motivator" model; alternatively, the first study may have reported an alpha error. To see whether the model has practical value, we recommend a study of landscape preference using participants with clinically significant levels of mood dysphoria.