The intensive use of the United Nations’ World Urbanisation Prospects database 1 (hereafter referred as WUP) belies growing concern in the scientific community about United Nations urban projection methods and results. Very few successful attempts have been made to offer a better alternative to the UN database that would be comparable in scope and regularity of publication. Certainly the national data sources that form the basis of the UN database are criticized (Hugo and Champion 2003), mainly for the data inconsistencies observed in less developed countries (LDC) and for the difficulty associated with the lack of comparable definitions of urbanization across countries (see article in this volume by Buettner). The quality of the data used for projections has a bearing on the choice of projection method, and therefore the paper begins with a brief overview of data sources and 1 The World Urbanization Prospects database refers to the publicly available estimates and projections for urban and rural, population and cities for locations greater than 750,000 persons, as described by Buettner in this volume. limitations. However, the weakness of urban projections does not principally lie on data quality but on modeling. The remaining of this paper will focus on shortcomings in the methodology and make suggestions for future directions in modeling. Dataorwhy weshould notwaitforthe better The lack of reliable data is a basic concern for urban population projections. The UN Population Division makes a strenuous and commendable effort to produce the WUP every two years. Because the publication cannot wait for all data to become available, it makes use of an inherently incomplete series. The WUP-2011 considers 2011 as its last observation year, but the last Spatial Demography 2014 2(2): In Press http://spatialdemography.org OPENACCESS via Creative Commons 3.0 ISSN 2164-7070 (online) RESEARCH Philippe Bocquier 1 1 Université catholique de Louvain, Belgium The Future of UrbanProjections: SuggestedImprovements onthe UNMethod ABSTRACT This paper argues that imperfect data should not hinder the search for alternative projection models and methods. Given the available data on urbanization, autoregressive projection models should be preferred to explanatory models for projections though explanatory models are more suitable for analyzing past trends. Alternatives to the UN’s projection method, Bayesian projection included, should account for the social and economic inequalities embedded in the global, national and sub-national urban systems. KEYWORDS: urban projections, projection methods, Bayesian models, world inequalities, urban systems Corresponding Author: Philippe Bocquier, Professor, Centre de recherche en démographie et sociétés IACCHOS, Place Montesquieu 1, bte L2.08.03, B-1348 Louvain-la-Neuve. Tel: 32 (0)10 47 41 83. Fax: 32 (0)10 47 29 52. Email: philippe.bocquier@uclouvain.be