Phylogeography is a field that has been constantly developing. Many methods exist that can be used to tease apart the demographic processes and events that affect a species evolution. One method in particular, Nested Clade Phylogeographic Analysis (NCPA), has become a popular method for doing this. It claims that the distance statistics it uses can discriminate between multiple demographic scenarios. However, a major limitation of NCPA is that it does not provide estimates of error. Lack of automation of two stages in NCPA has meant few datasets can be simulated. A fully automated package, ANeCA, is developed, tested and applied to published datasets to recover published inferences. Many inferences are recovered showing that ANeCA is representative of a typical application of NCPA. ANeCA is also applied to data simulated under a model of panmixia. Results show on average 76% of datasets gave rise to at least one inference of an event or process when there was no geographic association. This is due to the generation of multiple statistics per clade, and multiple clades per dataset. It is also shown that the statistics are not independent of each other, making it difficult to correct the multiple tests performed. ANeCA is also applied to simulated data under three models of gene flow; an island model; a stepping stone model; and a lat- , tice model with some long distance dispersal. The results show that NCPA has a tendency to infer Restricted Gene Flow with Isolation By Distance, and Contiguous Range Expansion, under panmixia and the three gene flow models. Unlike the panmictic model where other inferences were rare, many inferences are found in higher frequencies irrespective of the underlying gene flow model. Comparison between the island and stepping stone models show that NCPA is unable to discriminate between long and short distance movements. Three classical summary statistics; AMOVAj Fu's FSj and the Mantel test, were also calculated on each dataset. Results show false positives rates to be around 5% for the AMOVA method and Mantel test, and 3% for Fu's Fs at the deme level. In general the AMOVA method is more sensitive to population structure than NCPA, and the Mantel test is more sensitive to isolation by distance than NCPA. The results that have been obtained in this study suggest that previous publications that have used NCPA have most likely made false inferences and it is recommended that classical, and most probably model-based, methods of inference should be used in preference to NCPA.