As I've mentioned, one of the regular columns I really enjoy reading is the Wisconsin Soils Report. In the May/June 2013 issue, Doug Soldat asks "How reliable is soil testing?" He pointed out some of the limitations of soil testing – they are probably more extensive than you thought – and provided some good advice about practical and effective soil testing. The focus of the article is on the most common soil testing mistakes, and how to avoid them.
You will want to read the entire article, which you can download at the link, but in the meantime here are a few relevant quotes.
On soil test extractants: "While Mehlich-3 may not be the best test for all situations, it is regarded by many as the most versatile extractant and it’s the one we have the most calibration data for here in Wisconsin, with the Bray coming in a close second."
On soil test recommendations: "Turfgrass researchers continue to improve the soil testing recommendations, but that type of research is time consuming and expensive. It is also worth noting that every time a researcher conducts one of these studies, they tend to find that the levels required are lower than what we previously thought – meaning that “low potassium” you got on your last soil test report might be optimum down the road."
On recommended nutrient guildeines: "I recommend you compare your results with PACE Turf’s Minimum Levels for Sustainable Nutrition guidelines which can be found here: www.paceturf.org/PTRI/Documents/1202_ref.pdf ... the minimum levels published by PACE are drastically lower than many traditional soil test interpretations, and likely more accurate."