In a recent paper, (hereafter Dormann et al.) conducted a review of approaches to account for spatial autocorrelation in species distribution models. As the review was the first of its kind in the ecological literature it has the potential to be an important and influential source of information guiding research. Although many spatial autocovariance approaches may seem redundant in the spatial processes they reflect, seemingly subtle differences in approach can have major implications for the resulting description of the data and conclusions drawn. Though Dormann et al.'s review of the available approaches was a step in the right direction, we think that their simulation study ignored important concepts which leads us to question some of their conclusions.One of Dormann et al.'s primary conclusions was that parameter estimates for most spatial modeling techniques were not strongly biased except in the case of autocovariate models. In the autocovariate model, as implemented by Dormann et al., the parameter representing the effect of environmental variables on species distributions (the coefficient for rain) was consistently underestimated. For this reason Dormann et al. cautioned the use of autocovariate approaches. This caution reiterated findings from a similar simulation in which (Dormann 2007) argued that autocovariate logistic regression models used for binomially distributed data (autologistic models) would be biased and unreliable. These results appear to be in direct contrast to earlier evaluations of this method (Augustin et al. 1996, Hoeting et al. 2000, He et al. 2003) and need to be considered seriously, as autocovariate approaches are now widely used in ecology (Piorecky and Prescott 2006, Wintle and Bardos 2006, McPherson and Jetz 2007, van Teeffelen and Ovaskainen 2007, Miller et al. 2007); for instance the seminal paper on autologistic regression (Augustin et al. 1996) has now been cited 222 times (Web of Science accessed 8 September 2008). Simplified implementation and interpretation of these models may result in misleading conclusions.Our critique is on three grounds. First, we show that the change Dormann et al. observed in the parameter estimate between the autocovariate approach and the true value is due to multicollinearity between environment and space. Variation shared among parameters is a common occurrence in ecological models and can rarely be avoided (Graham 2003); however, it can be directly measured using hierarchical partitioning approaches (Chevan and Sutherland 1991). Second, there are situations in which autocovariate approaches offer the opportunity to incorporate effects of behavioural and population processes into ecological models. This may result in greater understanding of these processes even though interpretation of the estimated coefficients themselves may not be possible. Third, we highlight that statistical regression models are developed for different objectives than outlined by Dorman et al. In particular, the goal of predicting future or nonsampled observations ...