Is a nature lover automatically an environmentalist, or does the love of nature lead to environmentalism? Classical notions of attitude toward environmental protection and attitude toward nature do not distinguish between nature and environmental protection as two attitude objects. Implicitly these notions, thus, presume appreciation of nature and appreciation of environmental protection to fall into the same category. We argue that this is a misconception and propose a model in which the two attitudes are distinct concepts. In this model, we use attitude measures that are based on Campbell’s paradigm (see Kaiser et al., 2009) to test our hypothesis. Participants in our study (N = 1336) were mostly recruited through newspapers and Internet platforms. Their mean age was 30 with a female proportion of 44.8%. Our survey entailed three instruments: a classical environmental protection measure (the New Ecological Paradigm scale by Dunlap et al., 2009), a Campbell paradigm-based attitude towards environmental protection measure (the General Ecological Behaviour scale by Kaiser and Wilson, 2004) and a Campbell paradigm-based attitude toward nature measure (proposed by Brügger et al., 2009). While the first measure is a widely used traditional attitude measure, the last two measures were developed within Campbell’s paradigm, which states that attitudes are behavioural dispositions, and that behaviour and attitude are axiomatically rather than causally linked (Kaiser et al., 2009). As such, attitudes can be captured by inspecting people’s past behavioural performance. A multidimensional Rasch model (similar to a confirmatory Factor analysis) was implemented to test the anticipated itemfactor structure, specifically, whether the items are better represented by a two or a one-dimensional model. The model fit for the two-dimensional model was statistically better than the one for the one-dimensional model: D2(3) = 2’094.86, p < .001. However, when looking at reliabilities, both the two-dimensional and one-dimensional model yielded good results: rel. > .80 for all versions. Model accuracy seemed comparable as well: predicting actual responses with a onedimensional model was nearly as accurate as with a two-dimensional model: attitude towards nature p = .610 vs. p = .624, attitude towards environmental protection p = .686 vs. p = .695. Finally, all three measures correlated with each other substantially (ranging from r = .35 to r = .59; these correlations are corrected for measurement error attenuation). Our results indicate that technically, the two attitudes could be collapsed into a one dimensional model. Reliabilities and model accuracy gain only marginally with a two-dimensional model. However, since someone who appreciates nature is probably also more likely to favour environmental protection and vice versa, we had to expect a fairly oblique factor structure. Unsurprisingly, we also found correlated attitudes, which made the technical distinction progressively more demanding. From a practical point of view it, nevertheless, seems important to keep the two attitudes separate. Attitude towards nature might be malleable and, thus, a crucial factor when we aspire to promote more environmental engagement.