Purpose – This paper aims to make a preliminary estimate of the degree of integration in the US product market (widely acknowledged to be the most integrated among geographically large economies) as an upper bound of spatial integration that is practically achievable in markets covering fairly large territories. Design/methodology/approach – The approach takes the form of an econometric model derived from the fact that local price of a tradable good should not be dependent on local demand under the law of “one price is a tool to measure market integration”. It is applied to data on the cost of a grocery basket and prices for three individual goods in 2000 across 29 US cities. Findings – The regression results suggest that the US market is not perfectly integrated. Thus, the estimated degree of its integration can be deemed, indeed, as a feasible maximum. Applying this benchmark to the European part of Russia in 2000, its degree of market integration turns out to be comparable – by the order of magnitude – with the feasible one. The roles of a few factors that could potentially cause segmentation of the US market are estimated. Research limitations/implications – The estimated degree of US market integration is crude because of the relatively small spatial sample. Further research has to substantially widen the spatial sample and estimate integration of the US market across a number of points in time. Originality/value – The paper suggests a realistic benchmark standard for judging the extent of market integration in various (geographically large) economies.