In already drought-stressed areas and places with the potential for desertification as a result of greenhouse-induced change, high quality model-derived climate projections are essential for sustainable management. Today's challenge is how to select from the plethora of models and proposed new analyses the tools most likely to be valid for areas already water-stressed and those threatened by future surface moisture reduction. Here, the land-surface skills of models involved in the IPCC Fourth Assessment Report and new techniques of isotopic enrichment of components of evapotranspiration are analyzed. Both are found to have shortcomings. Surprisingly poor reporting of fundamental components of the land-surface system in standard model output was the largest challenge for widely accepted models. We show that very few of a large group (20) of today's climate models report land-surface water and energy budgets correctly in a well-controlled international experiment and that most fail basic conservation tests. Our analysis of a smaller (5) experiment suggests that isotopic techniques employed in arid zone irrigation management may not transition to evaluation and model improvement. Land-surface conditions important for policy are found to be poorly reported which raises questions about equal weighting given by international assessments to all models: good and bad.